LUND UNIVERSITY SCHOOL OF ECONOMICS AND MANAGEMENT

DIGITALIZATION AND FIRM PERFORMANCEAre Digitally Mature Firms Outperforming Their Peers?

a thesis presentedby

JULIAN B. WROBLEWSKIto

the department of economics

Supervisionby

PROFESSOR HOSSEIN ASGHARIAN

for the degree ofMASTER OF SCIENCE

in the subject offinance

June 2018

Digitalization and Firm Performance

ABSTRACT

This thesis studies the effect of digitalization on firm performance. More specif-

ically, it investigates whether digitally mature firms outperform their less digitally

mature peers. My analysis is based on literature on information technology, infor-

mation and digital systems, and digitalization in general. I assess digital maturity

through a suitable proxy. With data from listed Swedish firms, I deploy dynamic

panel data regressions, portfolio analyses, and risk factor models in order to analyze

digital maturity’s effect on subsequent operating performance and stock returns. For

the most part, I find inconclusive results. Consequently, the proposition of a digital

advantage is not reinforced. This adds to the literature by empirically examining the

effects of digitalization on firm performance.

Keywords: Digitalization, Information Technology, Portfolio Analysis, Dynamic

Panel Data Regression, Operating Performance, Risk Factor Models

II

Contents

1 Introduction 1

2 Theoretical Background 62.1 The Catalyst: Advances in Information Technology . . . . . . . . . . 62.2 Technology Trends and Data Capital . . . . . . . . . . . . . . . . . . 92.3 Digitalization’s Impact on Organizations . . . . . . . . . . . . . . . . 122.4 The Digital Disruption . . . . . . . . . . . . . . . . . . . . . . . . . . 152.5 Digital Transformation - A New Digital Future . . . . . . . . . . . . 19

2.5.1 Data Analytics . . . . . . . . . . . . . . . . . . . . . . . . . . 202.5.2 Computer Simulations . . . . . . . . . . . . . . . . . . . . . . 222.5.3 Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . 222.5.4 Digital Platforms . . . . . . . . . . . . . . . . . . . . . . . . . 232.5.5 Digital Transformation Drawbacks . . . . . . . . . . . . . . . 24

2.6 Digital Maturity and Firm Performance . . . . . . . . . . . . . . . . 252.6.1 Digital Maturity: A Framework . . . . . . . . . . . . . . . . . 262.6.2 Digital Maturity Effects on Firm Performance . . . . . . . . . 28

3 Methodology and Data 313.1 The Digital Maturity Measure . . . . . . . . . . . . . . . . . . . . . . 313.2 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353.3 Model Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

3.3.1 Portfolio Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 363.3.2 Risk Factor Models . . . . . . . . . . . . . . . . . . . . . . . . 383.3.3 Panel Data Regression . . . . . . . . . . . . . . . . . . . . . . 39

III

4 Empirical Research 424.1 Portfolio Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . 42

4.1.1 Digital Maturity and Portfolio Returns . . . . . . . . . . . . . 424.1.2 Digital Maturity and Risk Factor Models . . . . . . . . . . . . 45

4.2 Panel Data Regression Results . . . . . . . . . . . . . . . . . . . . . . 47

5 Discussion 505.1 Interpretation of the Results . . . . . . . . . . . . . . . . . . . . . . . 505.2 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

6 Conclusion 54

References 56

Appendix A 66

Appendix B 68

Appendix C 70

Appendix D 71

IV

List of Figures

1 The DI and TMI framework. . . . . . . . . . . . . . . . . . . . . . . . 272 Scatter plot of digital maturity levels for publicly traded firms . . . . 28

3 Digital maturity measure frequency histogram . . . . . . . . . . . . . 34

V

List of Tables

1 Sample data description. . . . . . . . . . . . . . . . . . . . . . . . . . 352 Pearson and Spearman correlation coefficients. . . . . . . . . . . . . . 36

3 Digital maturity and monthly average portfolio returns. . . . . . . . . 434 Results of the portfolio analysis . . . . . . . . . . . . . . . . . . . . . 445 Results Fama French three-factor model with DM portfolios. . . . . . 456 Results Carhart four-factor model with DM portfolios. . . . . . . . . 467 Results of dynamic panel data regression . . . . . . . . . . . . . . . . 48

A1 Appendix A: Sharpe ratio significance tests I . . . . . . . . . . . . . . 67A2 Appendix A: Sharpe ratio significance tests II . . . . . . . . . . . . . 67B1 Appendix B: Monthly size-adjusted portfolios excess returns . . . . . 68C1 Appendix C: Result of the risk-adjusted performance measure on DM

portfolios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70D1 Appendix D: Detailed results of dynamic panel data regression . . . . 71

VI

List of Abbreviations

DI Digital Intensity

DM Digital Maturity

GDP Gross Domestic Product

GMM Generalized Method of Moments

ICT Information and Communication Technology

IoT Internet of Things

IT Information Technology

SWIFT Society for Worldwide Interbank Financial Telecommunication

TMI Transformation Management Intensity

VII

Acknowledgments

This Master’s thesis completes a chapter of my life, but also lays thefoundation stone of the chapter to come. The process of writing a thesis can bechallenging and demanding at times, yet it is ultimately rewarding. I could not havedone this without the help of several people.

First and foremost, I wish to express my sincere gratitude to Professor HosseinAsgharian, my supervisor, for his patience, guidance, and council. I am aware that thewriting process of this thesis was everything but linear and, therefore, I am deeplythankful to have experienced Professor Asgharian’s understanding and, whenevernecessary, timely advice.

In addition, I would like to thank my good friends and former colleagues BerndA. Fürst and Dirk Heizmann for their advice and critical comments. Your input wasgreatly appreciated.

•

Most of all, I am truly grateful for the support of my loving parents throughoutmy years of study and especially on my journey abroad. I could not have done thiswithout you.

VIII

We always overestimate the change that will occur in thenext two years and underestimate the change that willoccur in the next ten.

William Henry Gates III

When my information changes, I alter my conclusions.

John M. Keynes, as cited in Samuelson’s The Keynes Centenary

1 | Introduction

Over the course of the past two decades, digitalization has turned from arather abstract and futuristic concept into a transformative power that reshapes theeconomic landscape, and acts as a major disruptor with far-reaching consequences(Weill and Woerner, 2015; Bughin and van Zeebroeck, 2017; Schwab, 2017; Zammutoet al., 2007).

It is hard to pinpoint the arrival of digital technologies on a timeline. A few first in-novations in information technology and the digital field cleared the way for countlesstechnological innovations through a cascading effect and defined today’s fifth Kon-dratieff wave1: the age of information (Vogelsang, 2010). Originally, the notion of theso-called digital divide was coined sometime during the rise of information technol-ogy in the 1990s. It was used to explain the differences in information inequality, i.e.using computers or the Internet altogether (Yu, 2006). Over time, progress in infor-mation technology made computers, online interaction and interconnectedness com-monplace (Yoo, 2010). The modern-day understanding of digitalization shifted awayfrom simply working with a computer to a consensus that digital technologies trans-form how businesses interact in both business-to-business and business-to-customersettings (Loonam et al., 2018).

Digital transformation is not just limited to tech-savvy start-ups and companiesoperating in the high-tech sector, established companies embark on this digital trans-formation journey just as well (Weill and Woerner, 2018). Digitalization representsa paradigm shift that pervasively affects the most traditional of businesses and evenimpacts society as a whole (Gimpel and Röglinger, 2015; Sambamurthy et al., 2003).

1Kondratieff waves are classical theoretical economic cycles that follow apparent oscillations inglobal activity (Korotayev et al., 2011). However, the validity of the concept is not without contro-versy (Focacci, 2017).

1

Historically, firms and organizations have always been subject to constant change. AsSchumpeter (1943) so aptly put it with his concept of creative destruction, changeshould be seen as an inevitability, though, as a positive one in an economic landscape,as it fuels innovation, competition and higher standards that benefit all market ac-tors. Nowadays, with the digital transformation in full swing (Bharadwaj et al., 2013),firms recognize their peculiar situation amongst a great many technological innova-tions that reshape the economic and competitive landscape (Loebbecke and Picot,2015). Competitive dynamics have picked up in pace due to advances in informationtechnologies and new digital solutions (McAfee and Brynjolfsson, 2008). This leads toincreasing gaps between leaders and laggards and thus, sectors more closely resemblewinner-take-all markets.2

Companies at the forefront of new innovative concepts, the so-called digital fron-tier, increasingly benefit from their practices as they are getting propelled onto newcompetitive levels from where they can increase their profits (Scott et al., 2017).The unprecedented growth of highly digital and innovative companies, regardless ofsize, and their ability to take market shares by storm, are a prime example of howswiftly the business world can change and how easily industry boundaries becomeblurred (Yoo et al., 2010). Established enterprises revamp their business strategiesto stay afloat in the digital economy (Weill and Woerner, 2018). In other words, theemergence of digital technologies puts unparalleled pressure on companies to evolve.3

Ultimately, companies are confronted with a digital imperative: adapt to the newstate of the art or risk competitive obsolescence (Fitzgerald et al., 2013).

It even seems that, due to the changes induced by digital technologies, establishedbusiness principles no longer hold. According to the traditional technology adoptionmodel by Rogers (1963), new innovations or products gain popularity sequentiallywith market segments, starting from innovators and early adopters to the late ma-jority and laggards. In comparison, the disruption caused by digital innovations ismore compressed where new products, such as smartphone apps or e-commerce, areimbibed swiftly by the vast majority (Downes and Nunes, 2013). This fast-pacedchange, stemming from disruptive technologies, requires companies to be prepared

2It is worth mentioning that however powerful the magnitude of information technology advancesand digital innovations seem to be like, they do not affect sectors equally.

3Based on data from Standard & Poor’s, Desmet et al. (2015) observe that the average corporatelife span is decreasing. They find that, although the average corporate life span has been falling forthe past decades, the dwindling has never been more accelerated.

2

for technology innovations and to react timely, if necessary. Additionally, the digitaltransformation process encompasses an unprecedented diffusion of digital disruptors,i.e. relatively small and young but highly innovative firms that abruptly compete withestablished companies (Schwab, 2017).

Digitalization and its implied changes do not only affect the way firms operate,but they also have a profound impact on consumers (Brynjolfsson and McAfee, 2014).An ever-growing number of business people and consumers act online since mobileand internet services are becoming more widespread available (World Economic Fo-rum, 2016). Be it for online purchases, leisure activities such as consuming media,or search inquiries for information, users leave a data trail which can be analyzedand exploited by firms (Bharadwaj et al., 2013). More and more, firms see data as aform of capital and, therefore, as an asset (Westerman and Bonnet, 2015). The datacompanies gather can be thoroughly analyzed and used to improve understandingcustomer needs. Ultimately, this new information provides the basis for computermodeling and forecasting which, for instance, allows for better managerial decisionmaking or dynamic pricing, amongst other things (Chen et al., 2016).

Gassmann et al. (2014) define digitization as “ability to turn existing productsor services into digital variants, and thus offer advantages over tangible products,e.g. easier and faster distribution” (p. 6). Strictly speaking, digitization only refers tothe technical aspect of the conversion from analog to digital (see OED Online, 2017;Yoo, 2010). On the other hand, the concept of digitalization goes further and de-scribes the use of digital innovations and technologies to enhance, for instance, inter-nal processes and business models. In short, the utilization of digital resources and ITtechnologies by using resources that were created by digitization. For further clarifi-cation, firms utilizing new IT innovations and digital solutions are said to be digitallymature. However, there is ambiguity when it comes to digitization and digitalizationin the literature and the terms are often used interchangeably and synonymously.4

The goal of this thesis is to investigate whether firms which are more digitally ma-ture have higher performance than peers which are not as digitally mature. I criticallyexamine if digital maturity has observable effects on key performance measures and ifstock prices incorporate this information. Since a plethora of papers and studies arguein favor of digitalization, the key hypothesis is that digital maturity is seen to be posi-

4This thesis’ intention is not to enhance the terminology, thus, I will exclusively use the termdigitalization for the remainder of the thesis.

3

tively associated with subsequent operating performance and abnormal stock returns.Furthermore, digitalization is a widespread concept and cannot easily be describedas a single item since it is more of an ongoing process. Hence, identifying the keyareas of digitalization is an important part of this thesis.5 In order to pinpoint majordigital drivers, the thesis will give a succinct, yet in-depth, overview of the followingtopics. First, this thesis aims to give an outline of digitalization, digital maturity,IT innovations as well as digital progress. Second, changing economic fundamentalsin the wake of digitalization are elucidated. Third, general digital developments areillustrated. This serves to give a concise overview of digitalization in a broader senseand helps to define this thesis’ goal in a framework.

After arriving at a definition of digitalization and its attributes, the thesis ties thetheoretical aspects of digital maturity and their effects on firm performance together.For empirical testing, I deploy dynamic panel data regressions, portfolio analyses,and risk factor models. I run the dynamic panel data regression with a GeneralizedMethod of Moments estimator suggested by Arellano and Bond (1991). For the port-folio analysis, I sort firms into two size groups and rank them according to theirattached digital maturity value. Next, I calculate the monthly excess returns of thesize-adjusted portfolio returns. In addition, I make use of the Fama and French (1993)three-factor and Carhart (1997) four-factor models to gain additional insights. My ap-proach draws loosely from the analysis of innovative efficiency and contemporary firmperformance by Hirshleifer et al. (2013).

The results of this thesis demonstrate that, on average, digital maturity is neitherpositively, nor strongly related to a firm’s operating performance. There are a greatmany reasons why digitally mature firms should outperform their less digitally maturepeers, or in other words, why digitalization should give rise to above-average perfor-mance, however, the data suggests otherwise. The results from both, the portfolioanalysis and the dynamic panel data regression, show that there is no clear relationbetween digitalization and firm performance. More specifically, the portfolio analysisgives somewhat inconclusive results. With regard to excess returns, there is little dif-ference between portfolios consisting of highly digitally mature firms and less digitallymature ones. At the same time, the portfolio returns of highly digitally mature firmsseem less volatile compared to returns of the portfolio of less digitally mature firms.

5Because of the many facets that are touched upon through digital innovations, it is needless tosay that not all of them will be acknowledged.

4

As for the dynamic panel data regression, digital maturity has a weak negative effecton subsequent firm operating performance which is in contrast to prior studies andfindings. However, some argue that weak or negative results might be observed owingto implementation processes as well as the very nature of technological innovationswhich take some time to unfold their full potential (Scott et al., 2017). In short,my findings do not substantiate the claim that digital maturity is linked to superiorperformance.

The remainder of this paper is organized as follows. In section 2, I provide anin-depth overview of related research, in particular, literature related to digital in-novations. In addition, the differing effects of digital innovations on organizationsare illustrated. This theoretical background also presents a suggested framework fordigitalization and succinctly highlights findings from previous studies. In section 3, Idescribe the setup of my study, the data, and the tools used in order to qualitativelyexplore the effects of digitalization on firm performance. Section 4 presents the em-pirical results from the dynamic panel data regression and the portfolio analysis indetail. Section 5 discusses the implications of the findings in the light of the reviewedliterature and studies. Additionally, several limitations to my approach are outlined.Section 6 summarizes the main findings and provides a direction for further research.

5

2 | Theoretical Background

This chapter offers an in-depth literature overview which consists ofthe following. First, an introduction about the changes in competitive dynamics dueto advancements in information technology (IT) is given. Second, this chapter shedslight on the effects of new technological and digital innovations on organizations: thedigital transformation. Third, a framework to classify digital maturity is presentedand discussed. Lastly, the chapter aims to answer the question why digitalizationmatters for firms by highlighting findings from previous studies.

2.1 The Catalyst: Advances in Information Technology

Digitalization in itself is not a singular movement. It is is an on-going process thatdoes not have an ending in a traditional sense and is fueled by accelerating computingpower (Evans et al., 2006; Vogelsang, 2010). The quintessence for the changes of theeconomic landscape is that advances in IT caused a drastic acceleration of digital in-novations which is, by some, labeled as a disruption (Bughin and van Zeebroeck, 2017;Downes and Nunes, 2013). Especially technology companies seem to grow exponen-tially and old business principles no longer seem to hold.6 Digitalization does notonly affect firms that are doing business in the technology sector. IT solutions alsoenhance all organizations’ internal operations; changes that are invisible to the nakedeye.

Information has always been important throughout human history. However, ithas never been so easy to collect, to store, and to analyze large quantities of informa-tion as it is today. Digitalization affects every organization and industry and yet it

6The valuation of small start-ups, especially the ones located in Silicon Valley, cannot be explainedby traditional performance measures (Damodaran, 2014).

6

is incredibly difficult to pin down what exactly this phenomenon comprises. Technol-ogy advances reshape the economy, businesses reinvent operations and come up withcompletely new products, and customer as well as supplier relationships, are deep-ened. Digital maturity powered by IT applications has become a defining element incorporate competition (Manyika et al., 2015).

Technological changes at many fronts simultaneously, a new competitive dynamic,and digital innovations which push the boundaries, disrupt traditional power struc-tures. McAfee and Brynjolfsson (2008) find that the rivalry in an industry becomesmore fast-paced and dynamic due to successfully implemented IT systems. Whilefirms which fail to adapt properly and timely risk to fall behind and become irrele-vant, first-movers pull away from the average company. In other words, the compet-itive landscape is altered by innovative enterprise IT (Yoo et al., 2010). Innovationsspread rapidly amongst firms and competitors do not only try to keep up, but alsotry to implement IT-based innovations that are already one step ahead. With theimplementation of new powerful information and communication information (ICT)systems, organizational structures are redefined (Loebbecke and Picot, 2015; Hylvinget al., 2012). Especially work processes that were previously limited by either worktime, office location or hierarchical work structures, are transformed into more flexibleand network-like structures (Zammuto et al., 2007). Process-centered organizationaldesigns, integrated databases, and technological systems form a relationship betweenorganization and ICT (Hylving et al., 2012). This allows companies to visualize workprocesses, use real-time analyses, engage in virtual collaborations, and run advancedcomputer simulations (Zammuto et al., 2007). Thus, ICT improvements are posi-tively linked to the economic success of organizations (Hernaus et al., 2012; Spanoset al., 2002).

Moreover, McAfee and Brynjolfsson (2008) argue that investments in IT are adriving force behind increasing competitive dynamics. Technology has changed tradi-tional competition and gaps between leaders and laggards widen (Schwab, 2017). Aconsequence is the concentration of sectors to an extent where they resemble winner-take-all markets. This constellation benefits firms that move quickly and before theircompetition. Lieberman and Montgomery (1988) find that technological leadershipis one of the main contributors to first-mover advantages and such advantages oftentranslate to enhanced future profitability. In more detail, the first-mover advantagedescribes the ability to earn abnormal economic profits - typically exceeding the cost of

7

capital (Kerin et al., 1992). This technological leadership is linked to digital maturitysince digitalization is a part of technological dominance. Missing a new technologicalinnovation such as a new digital application could have far-reaching implications for afirm’s future competitive environment (Fitzgerald et al., 2013). Furthermore, McAfeeand Brynjolfsson (2008) identify patterns in their data which support their argumentthat, due to frequent competitive moves, through which only the best companies cankeep up, competition becomes more concentrated and accelerated. IT can be both, acatalyst for new ideas and the means to implement them successfully. A major USretailer, for instance, records and stores data gathered from customers indefinitely.The aforementioned customer data, or shopper history, is obtained throughout theentire shopping experience, item for item. The company then maps and updates thedata by store and region. Computer models fed with customer data are increasinglyused to supplement managerial decision-making. Such knowledge is considered to bepowerful since the superior decision-making is reflected in profits (Hays, 2004).

In a study of the insurance market, Barrett and Walsham (1999) investigate theimplications and effects of the introduction of interorganizational systems and elec-tronic data interchange for transactions in the London insurance market network.In order to increase the market efficiency, an electronic placing support system wasintroduced. Barrett and Walsham (1999) find that these changes allowed brokersand underwriters to work in an electronically supported trading environment whichhelped to increase both profitability and efficiency. Even the classic role of a bank asa financial intermediary is challenged since start-ups are now able to perform tasksthat were solely in the domain of huge corporations before the age of digitalization(Bank for International Settlements, 2002). Scott et al. (2017) take the same line andstate that technological advances and new digital innovations, such as improved andmore powerful applications and systems, override current market conditions. Accord-ing to Schumpeter (1943), innovations then fuel the motor of creative destruction,where companies, which are the most innovative, increase their market share to thedetriment of less innovative competitors. In short, companies at the digital frontierincreasingly benefit from their innovations as they are getting propelled into newcompetitive positions from where they can increase their profits. Subsequently, themagnitude of the contribution of digital maturity is dependent on economies andsectors (Scott et al., 2017).

Evidence from the financial sector suggests that the incorporation of improved ICT

8

has major implications for the financial system. These implications are far-reachingand have altered the global financial system. This leads to the creation of newly in-vented financial products (Barrett and Walsham, 1999), better service, and a moreinterconnected global financial market. The following example might illustrate thischange more coherently. In a study of the impacts of global information processingservices (SWIFT) on the banking sector, Scott et al. (2017) find robust evidencethat technology investments have a positive and significant impact on profitabilityand performance in the long run. Despite the long-term benefits, they find that inthe short run a weak or even negative result might be observed. This can, on theone hand, be explained by in-depth organizational changes due to the implementa-tion processes of information services. On the other hand, weak early results can beexplained by the very nature of technological innovations, which usually take sometime to unfold their full potential. Additionally, Scott et al. (2017) find that bankswere pressured to implement the new technology due to competitors upgrading theirsystems.7 Inaction or waiting too long might put a smaller bank out of business be-cause it gradually worsens its competitive position. Yet, if they adapt in time, theirrange of potential clients significantly increases and thus, the change is ultimatelybeneficial. This example depicts the inevitable need for organizations to keep up withtechnological innovations (see also Manyika et al., 2015).

It is, however, difficult to pin down the exact impact of new technologies andsubsequently findings are hard to generalize due to the overall complexity of this topic.Yet, several studies come to the conclusion that ICT boost returns, increase firm valueand have a positive impact on performance (Jun, 2008; Anderson et al., 2006).

2.2 Technology Trends and Data Capital

Nowadays, mobile and internet services are widespread and increasingly available.According to the World Economic Forum (2016), there are more mobile devices thanhumans, and consumers’ online purchases are ever increasing. In the first quarter of2014, for instance, almost 80 percent of US consumers purchased something online.Moreover, consuming media digitally is on the rise. In the US, around 50 percentof media viewed or listened to was digital (World Economic Forum, 2016). Using

7One of the original SWIFT engineers said that any bank which is not on SWIFT loses businessbecause banks that have already upgraded will not do business without SWIFT anymore (see Scottet al., 2017, p. 996).

9

online services generates data trails that consumers leave behind. Consumer dataencompasses a variety of information that can be analyzed and subsequently exploitedby companies. This new insight thus provides a valuable source for adding value andreceiving feedback. For example, Netflix, a US-based streaming service, gains a lot ofview time and therefore higher retention due to its well-optimized recommendationalgorithm (Vanderbilt, 2013). As a result, people use the streaming service longerand are more willing to keep their current subscription.8 The World Economic Forum(2016) defines four key improvements for organizations that use data as an asset.First, in-depth insight into customer behavior and preferences allow firms to makemarketing more personalized. Second, digital platforms are a means to connect withcustomers and potential customers to increase brand strength.9 Third, product designprocesses can be improved (see also Hylving et al., 2012). Lastly, consumer data canbe monetized by companies by sharing with other stakeholders. All in all, these keyareas are estimated to influence both revenue and cost savings positively.

In this day and age, firms are in the midst of the change from computing archi-tecture, such as hardware as a capital asset, to data as a new form of capital. Datais considered an asset that companies become more aware of since the technologi-cal developments have implications on firms’ strategies, characteristics, and businessopportunities (MIT Technology Review Custom, 2016; Weill and Woerner, 2018). Inaddition, Bharadwaj et al. (2013) argue that the data excavated by digitalizationis effectively transforming the availability of information. Firms therefore no longeroperate under information scarcity but, with all the data readily available, enter atime of information abundance. Westerman and Bonnet (2015) add, in comparison tofinancial and human capital, data may be considered an asset class. Generally speak-ing, the term capital is subject to change throughout economic history. Typically,in classical economics, capital is any good that helps create another good or a ser-vice (Hodgson, 2014). Data capital, however, is more complex, harder to access andsometimes not obvious. Companies gather data from customers, but fail to analyze itproperly. Without adequate applications and an up-to-date computing infrastructure,some information that is potentially valuable for the company, cannot be extracted

8Other prominent examples, where customized advertising that is specifically targeted at indi-viduals is common practice, would be Facebook or Google. Also, for Amazon, the related articlesuggestions, created by consumer data analyses, are at the core of their business model (WorldEconomic Forum, 2016). See also chapter 2.3.

9This effect, however, also works against companies’ efforts to increase sales since consumers haveaccess to more information and a greater variety of products to choose from (Dehning et al., 2003).

10

from the data.10 In the MIT Technology Review Custom (2016) data capital is definedas follows (see also Vogelsang, 2010; Quah, 2003).

First, data is non-rivalrous which means that a single piece of data can be usedby different applications or algorithms simultaneously. This is in contrast to financialcapital. For instance, an unspecified amount of money can only be invested in asmuch as one opportunity. Second, data is non-fungible which means that certain datacannot be substituted for another one due to the uniqueness of each data piece. Third,data is considered to be an experience good. The usefulness of information containedin data is not immediately obvious. Only after analyzing data its true value becomesvisible. This includes a certain degree of uncertainty due to the following reasons.On the one hand, the information might be useless for the recipient, and working oranalyzing the data, therefore, is a waste of money and time. On the other hand, theinformation might be valuable, but the applications and algorithms used to extractand structure the information might not be able to do so. Information as a resource isboth used and produced on a daily basis for companies. Therefore, datafication11 anddigitalization of more internal activities have effects on a firm’s competitive strategy.

The MIT Technology Review Custom (2016) abstracts data capital with the fol-lowing three principles. First, data derives from activities. Almost all activities bycompanies can produce digital data if the companies decide to use some kind of de-vice or application in order to capture it. It is needless to say that not all activitiesprovide equally important data for the company. Therefore, it is of importance tofocus on activities that are linked to a company’s competitive advantage, revenue,and cost drivers as well as interactions with customers or partners. Second, datatends to increase the amount of data by implication (Yoo et al., 2010). Data analyses,data interpretations, and data-driven algorithms tend to increase the total amountof data over time.12 Dynamic pricing algorithms, for instance, that change pricingautomatically and dynamically when demand changes, profit from previously gath-ered data. The more data about demand changes is available for the computation –favorably with different conditions – the better the fine-tuning of the algorithm and

10Especially demographic data is becoming more important for numerous companies. It helps com-panies to understand clients and allows, for instance, advertisement campaigns to be targeted at themain buyer group directly (Westerman and Bonnet, 2015; MIT Technology Review Custom, 2016).

11A term first introduced by Cukier and Mayer-Schönberger (2014) in ‘Big Data: A RevolutionThat Will Transform How We Live, Work, and Think’.

12In a recent study about data valuation, Short and Todd (2017) find that companies’ data in-crease, on average, by 40 percent per year.

11

thus an improvement in future performance. This is helpful for firms, for instance,to adjust prices continuously or to increase the accurateness of credit approvals forcredit companies. Third and last, digital platforms tend to win. This is closely linkedto the winner-takes-all characteristics of new digital innovations. In other words, net-work effects are pivotal for digital services (Evans et al., 2006). According to Shapiroand Varian (1999), the network’s value is a function of its size. This means that themore firms are actively adopting the new standards, the more valuable the networkbecomes for other firms. Since firms profit from the network, they create affirmatoryfeedback effects and thus, make the network more attractive to others which in turnincreases the network’s size and benefits (Shapiro and Varian, 1999).

However, the inaccessibility of data assets leads to increasingly difficult firm val-uations. In 2016, when Microsoft Corp. acquired LinkedIn, a social network servicefor professionals, many wondered why the software giant was willing to pay USD 26billion (Short and Todd, 2017). It can be assumed that Microsoft Corp.’s main in-terest in the acquisition of this online professional network was LinkedIn’s user data.Valuation of data as it turns out, is highly complex and context-dependent. For in-stance, the value of data for an outside party might be drastically different from aninsider’s point of view. Monetary value from data can be derived in two ways forfirms. On the one hand, data can generate value indirectly. By using data internally,customer behavior can be explained and thus processes can be redefined. This leadsto an increase in profits. On the other hand, a company can generate monetary valuedirectly by selling or trading customer data.13

In summary, an in-depth analysis of the data available to a company may givethe company unique insight into various business activities that have not yet beentouched upon. For the majority of firms, the data available from analytics is the singlebiggest asset that they possess. However, some firms do not take advantage of thator are not aware of the usefulness of the information that data provides them with.

2.3 Digitalization’s Impact on Organizations

As companies of any sector invest in digital solutions and start digital initiatives,competitors are forced to act (Fitzgerald et al., 2013). Manyika et al. (2015) argue that

13The consulting company PricewaterhouseCoopers estimates the revenue generated from com-mercializing data in the financial sector to be USD 300 billion by 2018 (see Short and Todd, 2017,p. 18).

12

digitalization is inevitable for organizations and in order to become digitally mature,going digital by implementing new systems and purchasing IT systems is not enough.Considering that early movers typically realize a major payoff, a timely reaction ispivotal (Lieberman and Montgomery, 1988). As more and more companies choose togo digital and a critical mass is reached, a ripple effect takes place which leads to a newcompetitive dynamic and alters the commercial ecosystem. Digitalization can be seenas a collecting basin where new innovations are captured. Predominantly, this includesdigital applications which heavily rely on data, that affect how firms are operating aswell as how they structure and design internal processes. Although these changes donot affect all sectors evenly, none is unaffected by it (Schwab, 2017). Firms mainlyacting as intermediaries, for instance, are replaced by digital platforms and slowly putout of business. Manyika et al. (2015) and Yoo et al. (2010) find that digitalizationis a disruptor, but at the same time opens up rigid barriers which leads to newopportunities. They state that digitalization allows companies to increase operationefficiency, to broaden their innovation efforts, and to better allocate their resources.Fitzgerald et al. (2013) emphasize that today’s organizations have to reinvent the waythey do business constantly, otherwise, they risk falling behind.

Fuentelsaz et al. (2009) argue that the implementation of new technology directlyaffects firm productivity through changes in the production process. This increase inproductivity, or productivity effect, can be decomposed into two areas. On the onehand, technological change enhances productivity directly on a technical level. Onthe other hand, the operating efficiency altogether increases, however, this effect isharder to determine because of the variety of factors affecting it.

For the increase in operational efficiency, Manyika et al. (2015) state that compa-nies profit from e-commerce, cost savings, and streamlining of operations. The digitalretail giant Amazon, for instance, uses advanced algorithms that show customers re-lated products based on predictive computations and alters prices dynamically inorder to increase sales and profit (Chen et al., 2016). Moreover, retail banks use au-tomated digital systems, such as mobile channels and online presence, in order toincrease paperless work flows and to cut costs (Gordon et al., 2013). Furthermore,Manyika et al. (2015) argue that digitalization does not stop at the implementationof new technologies that automate processes and lead to substantial cost savings.Additional insight, such as analytics, helps firms to understand their customers bet-ter. A recent example from the automotive industry illustrates how customer service

13

may go beyond physical limitations. Tesla Inc., a US-based based car manufacturer,can update the software of its electric cars without the assistance of the car’s owner(Kessler, 2015). These updates are downloaded automatically and improve currentfeatures or add new ones.14 As mentioned earlier, new technological innovations, so-lutions, and processes speed up the competition and may overwhelm individuals.Digitalization disrupts traditional companies’ daily business and operations. How-ever, a new hyper-competitive world fueled by digital transformation offers smallerbusinesses new opportunities since the barriers of entry are lower than ever. Smallbusinesses have fewer constraints, are typically more innovative, and are unburdenedby outdated legacy systems (Manyika et al., 2015).

According to a study conducted by Dobbs et al. (2015), firm profits for idea-intensive sectors are rising, especially leading firms in sectors with a higher averagedigital maturity are gaining market shares. This new competitive landscape influ-ences the modus operandi for established companies’ M&A activities, especially inhigh-tech or highly digitalized sectors. Smaller merger and acquisition endeavors thatwould not have been undertaken with regard to traditional valuation practices aresuddenly on the rise (Picker, 2017). More specifically, smaller firms with a lot ofintellectual property and cutting-edge technologies are sought after by larger compa-nies which try to either take over the high-value intellectual property from the smallfirm, or to thwart off competitors before they turn into a real threat (Hildebrandtet al., 2015; Luckerson, 2015). The Achilles’ heel, however, for a M&A transactionin high-tech sectors to be beneficial, is the proper integration of the acquired firm’sknowledge into the acquiring firm (Cloodt et al., 2006). Moreover, for the acquisitionto generate long term benefits in innovative performance, the firms involved in theM&A transaction should be somewhat related but not too similar with regard to theirindividual knowledge base (Cloodt et al., 2006).

Van Bommel et al. (2014) find that virtual environments, the ubiquity of big data,and digital channels change the consumer decision journey. Today, consumers knowmore about products and are aware of their enhanced choice. Lucas et al. (2013) labelthis increased knowledge of products as consumer informedness. Hence, it is importantfor companies to stay in touch with customers that use mobile phones to search fornew products and ultimately purchase them. Van Bommel et al. (2014) remark that

14In late 2015, with an overnight update, Tesla Inc. updated their car software by adding advancedself-driving capabilities (Kessler, 2015).

14

companies not only need to capture data but should also use sophisticated analyticsin order to interpret the data. In marketing, for instance, algorithms model costs,determine the efficiency of a new campaign, and identify new customer segments.

Furthermore, Manyika et al. (2015) investigate potential future economic growthdue to changes induced by the ongoing process of digitalization. More specifically, theyanalyze two topics in-depth which they consider to have an impact on future growth,namely, capital efficiency and multifactor productivity. On the one hand, the Internetof Things (IoT) offers great potential to use assets more efficiently and effectively,thus business processes increasingly revolve around data-driven services (Pflaum andGölzer, 2018). Automated systems in manufacturing companies, for instance, help toreduce outages and downtimes, and to increase throughput through the analysis ofperformance data (Haverkort and Zimmermann, 2017). With the help of sophisticatedIT and IoT applications and systems, there are no limitations for improving workflows, streamlining operations and cost reductions (Manyika et al., 2015). On theother hand, gathering information in digital form and subsequently analyzing data,improves management decisions and operational efficiency. For instance, real-timedata on inventory and forecasts of demand, computed by predictive algorithms, helpcompanies to decrease mismatches between the two and therefore mitigate both stock-outs and excess ordering. Advanced software allows for more realistic simulations andtherefore helps designers or engineers to use less raw materials and to eventuallyarrive at a leaner production approach. All in all, the impacts of digitalization onthese main areas are estimated to boost the annual gross domestic product (GDP) inthe US by approximately USD 1.6 trillion to USD 2.2 trillion by 2025.15

2.4 The Digital Disruption

The rapid digitalization of the business world causes turmoil. Traditional industrybarriers are broken down, which results in lower barriers of entry for smaller busi-nesses, such as start-ups, which are generally highly innovative and more flexibledue to fewer legacy systems. Digitalization as a disruption has both the potential tocreate new business opportunities and the power to destroy long-successful businessmodels (Weill and Woerner, 2015). In this new digital ecosystem, traditional firms,regardless of how well-established they are, may struggle to keep up. New forms

15McKinsey Global Institute analysis (see Manyika et al., 2015, p. 61).

15

of customer relationships and engagement allow innovative businesses to overtakemarkets (Downes and Nunes, 2013) and engage in unique and unexplored businessopportunities (Loebbecke and Picot, 2015).

However, the digital era and its disruption is a double-edged sword. This meansthat companies that exploit the opportunities provided by digitalization thrive. Oth-ers that do not utilize new technologies fall behind. In a recent study, Bughin andvan Zeebroeck (2017) investigate the magnitude of the digital disruption’s impacton (1) companies, both traditional, that is typically less digitally mature ones, andnew digital entrants as well as (2) industries. They find that incumbents16 experiencea negative impact on profits due to the fierce competition caused by new entrants(see also McAfee and Brynjolfsson, 2008). In other words, digital entrants changethe dynamic of markets. Bughin and van Zeebroeck (2017) estimate that new digi-tal entrants take up to 20 percent of market shares. Moreover, new entrants do notonly seize shares of revenue but also change customer behavior which puts additionalpressure on traditional firms.17 Moreover, the threat of new entrants for incumbentsis not only the loss of market share, it is also the head start new entrants have iftheir strategy appears to be successful. Incumbents then have to react timely andadequately. Additionally, Bughin and van Zeebroeck (2017) find evidence that com-panies that try to tap their full digital potential gain the most and their returns arehigher compared to the average firm. While 90 percent of firms engage in digitaliza-tion efforts18, only a third of all companies try to conquer new markets which is oneof the big opportunities digitalization has to offer. The two key areas firms have tofocus on, are new customer segments and new markets. New customer segments arisefrom the use of a digital platform, or social media, in order to reach a new audience.New markets can be developed by selling differently. It is estimated that, on average,one third of firms’ revenues are digitized globally. As their main finding, Bughin andvan Zeebroeck (2017) note that, on a global scale, incumbents’ revenue and EBITgrowth is reduced by up to 30 and 25 percent, respectively, due to the effect of dig-ital disruption. This revenue reduction for traditional firms mainly stems from twomajor effects. The first effect manifests itself in the very existence of new entrants.

16Companies that have been around for a longer period of time, in other words, establishedcompanies.

17For instance, the navigation service provided by Google Maps took over the market for stand-alone GPS makers in less than two years (Downes and Nunes, 2014).

18However, less than ten percent of respondents in the study state that their business processesand operations are digitized on a high level.

16

As mentioned above, digital entrants claim market shares. New entrants, however,do not only take over market shares aggressively, they also increase the size of theindustry by an estimated 0.5 percent yearly. The change in market dynamics resultsin a new level of competitiveness and causes yet another change. Legacy companiesengage in the competition against both new entrants and other legacy companies.Bughin and van Zeebroeck (2017) label this as some sort of Red Queen19 competition(see also Dutta et al., 2014). When a company launches a new digital innovation,others must respond in a similar manner which might result in an unintentional racefor revenue-reducing moves.

The package-delivery market offers an illustration of this predicament. Over yearsthe two major US parcel services, FedEx and UPS, were in close competition forthe package-delivery market. As one of them introduced online parcel tracking, forinstance, the other reacted quickly and did not only add a similar feature shortlyafterward but also pushed other digital features in order to surpass the competitor.This led to higher spendings on IT than on transportation assets over time (Duttaet al., 2014). Normally, new value added by major investments are not realized in-stantaneously. Also, countermoves by competitors require some time during whichthe first-mover benefits. However, IT investments and new innovations increase thespeed with which companies react to each others’ moves. Therefore, it does not comeas a surprise that organizations invest heavily in IT to reduce the delays in com-petitive responses (Dutta et al., 2014). This is what strategy literature labels RedQueen theory or competition. Similarly, in financial services, as mobile payments andmobile banking became more widespread, traditional banks had to react, by eitherreducing or eliminating fees.20 Ultimately, both effects put pressure on incumbentfirms’ revenue and subsequently reduce margins (Bughin and van Zeebroeck, 2017).Analogously, there is a relation between the digital maturity of an industry and theseeffects mentioned above (McAfee and Brynjolfsson, 2008). They tend to be more se-vere for firms in highly digitalized sectors, particularly for companies that have notyet utilized their digital potential. In the high tech sector, for instance, the impacton revenue growth is four times the average compared to other sectors. However, in

19This is a reference to the famous sentence “it takes all the running you can do, to keep in thesame place” in Lewis Carroll’s (1871) Through the Looking Glass; closely related to the proverbialshark that needs to keep swimming in order to survive.

20The reduction or elimination of fees of one bank forces other banks to react, otherwise, consumersmight shift services. Needless to say, such actions are typically considered revenue-reducing moves.

17

manufacturing which is a less digital mature sector, the effect is attenuated to a mere60 percent of the average (Bughin and van Zeebroeck, 2017).

Digitalization is a central catalyst when it comes to the unprecedented levels ofcompetition next to other major factors that include higher M&A activities, moreR&D investments and local markets that open up to global ones.21 More specifically,IT investments and, correspondingly, more powerful IT systems are major driversfor the new competitive dynamic (Yoo et al., 2012). This new dynamic results in afiercer competition amongst organizations and an increasing gap between individualcompetitors, the leaders and laggards, in an industry. Ultimately, this ends in higherconcentrated industries where leaders try to revamp business in order to stay on thetop, whereas laggards are left behind. New technological innovations are at the core ofnew competitive dynamics (Haverkort and Zimmermann, 2017). In particular, digitalproducts and especially digital internal processes are accelerating competition. Forinstance, a company that reinvents its business processes and comes up with some-thing more unique and more efficient can easily implement this change in the entireorganization with high accurateness. Enterprise software and cutting-edge network-ing technologies help to implement the new processes easily and independently ofany physical restriction, such as a different geographical location. Thus, a valuableinnovation which allows a company to perform better has the potential to dominate asector. Direct competitors, in turn, try to roll out innovations also to pull ahead andgain market shares and thus the frequency of competitive moves and their responsesis increased (McAfee and Brynjolfsson, 2008).

Dutta et al. (2014) deploy a sophisticated model in order to investigate the me-chanics of competition with regard to digital systems and competitive responsiveness.In simple terms, they analyze the dynamic behavior with varying levels of investmentsin digital systems between companies. Their results indicate that digital innovationshelp to capture market share for both the innovative firms, the first-mover, and itscompetitors. Moreover, the first-mover may stay ahead of its competitors if the invest-ments in digital systems are substantial and do not fall below a certain threshold.22

However, if the first-mover fails to stay above the threshold, it runs the risk of getting21It is an ongoing dispute amongst researchers which of these shifts essentially drives the changes

in competitive dynamics. For instance, researchers find contradictory results when it comes to theimpact of R&D spending or M&A activities on competition changes (see Comin and Philippon, 2005;White, 2002; Cloodt et al., 2006). However, this assessment is beyond the scope of this thesis.

22The results also depend on the level of customer sensitivity.

18

surpassed by a competitor if the competitor chooses to invest more in digital systems.The threshold to dethrone a competitor varies with the intensity of customer sensi-tivity in a certain sector. As a result, returns from new investments in digital systemsare diminishing for firms (Dutta et al., 2014).

McAfee and Brynjolfsson (2008) identify a stronger link between technology ad-vancements and an increased competition today, compared to the levels in the 1990s.Additionally, McAfee and Brynjolfsson (2008) find evidence that the increase in thequality and quantity of IT investments has a direct impact on industry concentra-tion, turbulence in markets and performance spread.23 As expected, their findings arethe most pronounced in IT-intensive industries. Moreover, they contend that thesealterations in competitive dynamics are not yet completed as they are an on-goingprocess of organizations fostering and propagating process innovations.

2.5 Digital Transformation - A New Digital Future

Digital technologies encompass data analytics, big data, social media, cloud comput-ing and smart devices amongst other things. Traditional firms are slowly starting totransform their business due to several reasons. Generally speaking, a transforma-tion alters conventional ways of doing business fundamentally Dehning et al. (2003).For the digital transformation, two main concepts act as a watershed. On the onehand, firms acknowledge the potential of new technologies which have the power toredefine operational processes and business models (Yoo et al., 2010; Schwab, 2017).However, it is not as simple as implementing a new technology that drives techno-logical progression, it is the attentive and smart selection of advanced systems thathelp to redefine functions and push the boundaries of the firm (Westerman et al.,2011). On the other hand, firms feel the pressure from customers and competitorsalike (Fitzgerald et al., 2013). For customers, social media and online services areincreasing in importance ever since. If a company fails to satisfy customers’ needs, itmight face severe limitations for a successful future. Also, a new digital comparativeforces firms to either keep up with new techniques and processes or to face compet-itive obsolescence (Fitzgerald et al., 2013). In addition, organizations are starting tounderstand the power of data which is the fuel for data analytics.

23They find some correlations between M&A activity, R&D, globalization and the change of com-petitive dynamics, however, they argue that none of them were superior to their measures.

19

2.5.1 Data Analytics

Analytics is changing the way companies can transform insight into action. New tech-nologies allow companies to think ahead, i.e. use information gathered to improveforecasting. Knowing what happened and for what reasons is no longer sufficient.Organizations need to be kept informed about what is happening at the momentand what is likely to happen in the near future. These insights are superior to tra-ditional measures and metrics since they allow organizations to react timely andadequately whenever necessary (LaValle et al., 2011). Traditional decision supporttools and business intelligence are superseded by technical innovations and advance-ments in big data analytics. Typically, big data analytics refers to the “focus on verylarge, unstructured and fast-moving data” (Davenport, 2014, p. 10). Sophisticatedalgorithms, artificial intelligence and cognitive computing capabilities which improvedigital services and products, are dependent on big data analytics.

More precisely, digital transformation manifests itself in two key areas. The firstis the increase in computational power and big data which provide the basis forcomputer modeling, forecasting and artificial intelligence, amongst others (Zammutoet al., 2007). In short, the implementation of innovative digital solutions that allowcompanies to operate in a digital ecosystem. The second is the more in-depth in-sight into human reasoning and judgment. Organizations have the desire to knowmore about customers and customer behavior in general. Digitalization efforts typi-cally revolve around the integration of these two streams (Weill and Woerner, 2015;Schoemaker and Tetlock, 2017).

In order to assess the impact of advanced analytics, LaValle et al. (2011) conducteda global survey in collaboration with the IBM Institute for Business Value. Accordingto their findings, the extensive use of information and its interpretation with analytics,increases firm performance. Top-performing companies use, on average, five timesmore analytics than lower performers. Organizations want more insight, at a fasterpace and try to enhance their analytics approaches. Data management with strongabilities to capture, store, aggregate and analyze data separates organizations and actsas a differentiator. Managerial decision-making is not anymore only supplemented bybut based on data-driven decisions. Optimal solutions to a variety of scenarios arecalculated by computers through simulations, mathematical optimization and choicemodeling. LaValle et al. (2011) find that top performers rely more heavily and moreoften on business information and analytics. In other words, there is a clear trend to

20

rely on analytics rather than intuition. A common struggle companies face, however,is to find an adequate way to analyze the sheer amount of data. According to thestudy, the main issue is the lack of understanding how to make use of analytics. Thissupports findings that adoption barriers are mostly managerial (LaValle et al., 2011;Kane et al., 2015; Schoemaker and Tetlock, 2017).

Another aspect is the amount of data available today which is ever increasing.Companies actively collect and mine data from a multitude of internal and externalsources.24 Similarly, the cost of processing and storing data, even on a massive scale,is negligibly low. These effects allow companies to deploy machine-based readoutsand with the help of appropriate software, the interpretation of data in order to en-hance decision-making (Loebbecke and Picot, 2015). With a sheer abundance of dataand more sophisticated technical advancements in data analytics, organizations relymore heavily on data-driven managerial decision-making. Furthermore, Loebbeckeand Picot (2015) argue that digitalization is still reducing marginal costs and leadsto positive economies of scale25 which favors a more centralized production approach.

Ultimately, the goal is to optimize business processes, to increase efficiency and toenhance product quality as well as service quality powered by cutting-edge ICT sys-tems (Loebbecke and Picot, 2015; Majchrzak and Malhotra, 2013; Spanos et al., 2002).Big data analytics and data abundance allow firms to predict the present and to im-prove their forecast for the future based on information stored in the data. Firmsthat sell to a broad variety of customers, such as retailers, use big data analytics inorder to analyze buyer behavior. On the one hand, this allows firms to target certaincustomers directly through customized advertisements. On the other hand, discover-ing what customers usually buy together, or if a special event comes up, allows firmsto adjust their stock accordingly. For instance, Walmart, a US retailer, uses datafrom both weather and inventory stock in order to predict future sales. By analyzingshopper history inside Walmart’s computer network, the company starts predictingsales before they are going to happen (Hays, 2004). Both uses of customer behavioranalyses eventually increase sales and reduce mishaps caused by miscalculations.

24For instance, Google Trends, a publicly available report service provided by Google, shows thevolume of any search phrase listed ranked according to general search interest and based on regionalimportance. Most search queries are an indication of a person’s interest which can be exploited bydata analytics.

25Independent of the digital good, whether it be for entertainment (music or film) or businesssolutions (software or application), in short, intangible goods, can after completion be distributedglobally with minimal marginal costs.

21

Moreover, big data analytics does not have to stop at the customer side. Processesspanning over the entire supply chain can be stored and analyzed by computer models.The ultimate goal for firms is twofold: to increase efficiency at all levels throughoptimization and leaner operations and to use data in order to forecast and drive thebusiness (Hays, 2004).

2.5.2 Computer Simulations

Advanced computer models have the capability to run simulations and to conductwhat-if scenarios. Companies use dashboards that show data from business operationsin real time. In this way, users can act early and resolve problems when performanceindicators suggest that there may be a problem. For manufacturing companies withmultiple steps, this helps to avoid situations where later steps in the production linemay be affected. The full potential of using data unravels when it is being used forsimulations and forecasts. Algorithms and the use of artificial intelligence make itpossible to identify patterns in data. Findings are then projected in order to pre-dict buying behavior, supplier relationships, and competitor as well as industry-levelbehavior. What-if analyses allow firms to detach real figures and replace them withalternative actions to observe changes (Zammuto et al., 2007). This allows firms todiscover new possibilities such as alternative actions, to reduce uncertainty and topush into new directions. For instance, a manufacturer can run a simulation of whathappens to supplier availability if it decides to change certain parts or materials inan assembly process. In short, simulations augment organizational and managerialdecision-making (Zammuto et al., 2007).

2.5.3 Cloud Computing

Another topic which is part of the digital transformation is cloud computing. Calle-waert et al. (2009) state that using external computing power introduces unprece-dented opportunities for organizations that heavily rely on the aforementioned com-puting power. The characteristics of cloud computing are as follows. First, since thecloud primarily is an internet-based service provider that organizations can subscribeto, it is immediately scalable for an individual company. If a company wants to store asignificantly larger amount of data, it can simply increase the storage provided by thecloud with minimal time delay. Additional capacities are at the ready if the companywants more storage. Second, due to their high degree of abstraction, cloud services

22

are pay-as-you-go. This improves cost-effectiveness for organizations since they onlypay for services used. If an organization wants to run less analytics, it can offloadunused capacity quickly. Third, in case that an organization wants to use additionalanalytics, its own infrastructure which is physical hardware is no longer limiting thecomputational power that an organization has access to. This means that companiesare more flexible with regard to cloud services. More specifically, unexpected eventsno longer push organizations to their limits since they can easily scale up or downtheir cloud usage (Callewaert et al., 2009).

2.5.4 Digital Platforms

Digital platforms26 are a sublime example for making new markets accessible for firmsand targeting selective customers with the help from insights through data analyt-ics (Tiwana et al., 2010). For instance, Danske Bank, a Danish bank, revamped itsonline presence with an e-banking platform a couple of years ago which led to ben-eficial relationships with customers. As a result, nearly two out of three customersuse the online service for bank transactions (Weill and Woerner, 2018). For retail-ers, algorithm controlled recommendations of related products bring new productsto customers’ attention. Technically speaking, the recommendation list is put to-gether based on previously gathered data generated by the digital trails of customersand ultimately plotted as customer choice patterns. The more the company knowsabout the customer (based on the customer’s search history and past purchases), themore accurate and relevant the recommended product list will be (Oestreicher-Singeret al., 2017). However, Oestreicher-Singer et al. (2017) also argue that a company’spromotion of its products might as well benefit their competitors due to the digitalnetwork’s demand spillovers. It is difficult to anticipate the network structure effectsfor an individual company.

Yet another interesting insight is the recommendation network’s guidance whenit comes to products purchased together by customers. Companies might realize thatthere is a product offered by a competitor that is often bought together with theirown product, and thus recommendations reveal demand characteristics. This couldhelp companies upgrade their products or plan entirely new ones (Oestreicher-Singeret al., 2017). Numerically speaking, research conducted by the McKinsey Global In-

26This can be either a firm’s own web presence, other online services or the listing on retail websitesetc.

23

stitute (MGI, 2015) estimates that online platforms might add nearly USD 500 billionto the GDP by 2025 in the US.

Moreover, new developments such as crowdsourcing27 are surfacing. For instance,General Electric, a US energy giant, launched a sustainability challenge where out-siders were asked to suggest a solution to the issue at hand. In the wake of thisproject General Electric acquired a small firm and thus unconventionally added newbusiness intelligence. Essentially, crowdsourcing is a means for companies to ask cus-tomers what they should add to or improve in existing products. Firms benefit froman outpour of ideas and creativity, however, often suggestions are not feasible orimplementable in the end (Majchrzak and Malhotra, 2013).

2.5.5 Digital Transformation Drawbacks

The above-mentioned features and advantages paint a very optimistic picture of thedigital future. Yet, there are obstacles to circumvent for organizations and pitfalls totake into consideration. Using digital technologies is only one part of going digital. If acompany does not know how to interpret data or what it is looking for in specific, dataand analytics will not have a significant impact. Especially physical industries requirea sophisticated digital business strategy in order to adapt (Bharadwaj et al., 2013).Kane et al. (2015) argue that the greatest difference between digitally mature and lessmature companies comes from strategy, culture and the company’s vision of a moredigital future. Essentially, companies which are willing to use digital transformationand adjust the business accordingly are amongst the leaders in their industry. Thesystematical integration of computational power and the understanding of humanjudgment, or the human mind, lead to substantial competitive advantages (Schoe-maker and Tetlock, 2017). Most companies want to use digital technologies, but onlysome of them know how to use them to their advantage.

Kane et al. (2015) state that a firm’s organizational culture is crucial for theimplementation of digital technologies. This is also one factor why smaller businessespotentially profit more from digital innovations because they are less restricted with aleaner organizational structure. Additionally, smaller and fast growing enterprises canbuild their infrastructure with advanced ICT in mind, whereas established companies

27Crowdsourcing is defined as an online activity in which individuals or institutions can participatein order to arrive at a solution for a task presented by generally another organization. In other words,it is the process of an open announcement that invites outsiders to participate in a task (Estellés-Arolas and de Guevara, 2012; Gassmann et al., 2014).

24

have to deal with legacy systems and overlapping IT infrastructure. Centralized andstandardized systems increase agility in processes across the firm. The core platformused throughout the organization, processes integrated data and increases efficiency.Local variations of the standard platform can then be created for single divisions. If afirm fails to standardize digital processes, one result might be messy data and entailshigher long-term costs (Westerman and Bonnet, 2015).

According to Westerman (2017), the true value of technological innovations doesnot come from data analytics and more advanced algorithms, but from doing busi-ness differently. At its core, digitalization allows companies to understand internalprocesses and customers better. Although there is a necessity to engage in digital-ization (Schwab, 2017) or digital transformation, not all companies have embracedbecoming digital as part of a new corporate strategy. In a global survey of executivesand managers, conducted in 2015 by the MIT Sloan Management Review, the vastmajority (almost nine out of ten) of respondents agreed to the statement that digitalinnovations and digital transformations will shape their industries to a moderate orgreat extent. However, 56 percent of the respondents stated that their company isnot properly prepared for a possible digital disruption (Kane et al., 2016).

2.6 Digital Maturity and Firm Performance

In general, new digital technologies gain ground quickly and change the economiclandscape. Innovations put forth by small and fast moving innovative companies takeover industries and reshape how business is conducted. Traditional companies whichare typically more inflexible are faced with rapidly moving digital technologies thatpermeate industry standards. Digital innovations offer new opportunities, however,only a few companies capture the real benefits. The digital innovation level and affinityof companies can be captured in a metric: digital maturity.

Kane et al. (2016) argue that digital maturity in different organizations stemsfrom common features. Specifically, they identify the following features which drivedigital maturity. First, digitally advanced organizations have a higher risk tolerance.They accept a certain level of risk which, logically, is attached to the implementationof new technologies. Second, more digital mature organizations are willing to experi-ment which is closely linked to the higher risk tolerance. Rolling out new technology iscumbersome and might disrupt day to day business at first. Third, digitally maturing

25

organizations invest heavily in the recruitment of talents and leaders with transforma-tive visions. Those key employees help to shape a digital culture in the organization.It is important that digital innovations, organizational culture, and employees of anorganization are in sync.

Bughin et al. (2017) argue that a company’s future success in a more digital envi-ronment can be partially derived early on in its transformation journey by analyzingits digital intelligence. Digital intelligence28 is one of the key drivers to a successfuldigital transformation, besides, clear digital vision and adequate technology improve-ments, as previously mentioned (Kane et al., 2015). Furthermore, Bughin et al. (2017)find that digital intelligence is positively correlated with financial performance. In de-tail, a higher score of digital intelligence has a positive impact on revenue, EBIT, andcompany growth after controlling for industry, company size, and location. Their re-sults are statistically significant and the correlation holds in various industries, i.a.business services, manufacturing, and high-tech. Moreover, they find that (1) digitalintelligence scores are the highest for digitally savvy firms and (2) the effect of thefour dimensions (and their sub-dimensions) on digital intelligence is approximatelythe same, meaning the individual dimensions carry roughly the equal weight. Inter-estingly, the individual scores of the four dimensions are very consistent with thegroup ranking. In other words, digital leaders have higher scores across all dimen-sions compared to the average group and the same holds for the average group andlaggards.

2.6.1 Digital Maturity: A Framework

The following is a framework put forth by the Massachusetts Institute of Tech-nology Center for Digital Business (Westerman et al., 2011). According to this re-port, digital maturity can be seen as a combination of two dimensions. On theone hand, digital innovations which incorporate technology initiatives, new inter-nal processes, and business models. This is captured by digital intensity (DI): pro-cesses and new technologies that change how the company operates. On the otherhand, company-wide transformations which incorporate leadership changes, new vi-sions, and the mission to drive digital transformation. This is captured by trans-

28In their study, Bughin et al. (2017) use the following four dimensions to assess digital intelli-gence: strategy, culture, organization and capabilities. The capability dimension encompasses the ITarchitecture, whereas the first three describe the digital strategy and management practices. Theyobtain the scores by interviewing 250 firms around the globe.

26

formation management intensity (TMI): IT-business relationships and governancethat change the vision of the firm’s future. Figure 1 shows the proposed catego-rization of digital maturity. DI and TMI are divided into two sections that re-flect a company’s progression level (Westerman et al., 2011). Companies with lit-tle digital capabilities, that is, low DI and TMI levels are labeled as Beginners.

Figure 1: The DI and TMI frameworkand the corresponding categories (Westermanet al., 2012, p. 4.).

Beginners are either firms that are get-ting started or firms that are unawareof digital change and its advantages.Firms that implement new digital solu-tions, have a willingness to experiment,and strive for a digitally powered change,that is, high DI and low TMI are cate-gorized as Fashionistas. These firms usea brute force approach to become digi-tal leaders. However, they often miss thecrucial step of implementing proper gov-ernance internally, which dampens theoutcome of digital innovations becausedigital efforts cannot be utilized opti-mally. Digital Conservatives are compa-nies that value slow but steady progress

over a fast digital overhaul, that is, low DI and high TMI. These firms emphasizecorporate culture and good management, and they have a more skeptical outlookwhen it comes to digital innovations. Firms with high DI and high TMI are clas-sified as Digirati. Companies of this type are believed to fully understand how toutilize digital innovations and the implementation of such innovations. They excel atdriving value through digital transformation. They balance vision and engagementoptimally, invest in upcoming technologies, and develop an adequate digital culture.In general, companies at the digital frontier encompass characteristics such as a highspeed of innovation, automation of tasks, openness to experimentation, and the cre-ation of digital assets (Manyika et al., 2015). The balanced coordination of both,technologies that increase digital intensity and a digital road map that outlines firms’digital future, further strengthens their digital competitive advantage (Westermanand McAfee, 2012).

27

2.6.2 Digital Maturity Effects on Firm Performance

The study conducted by Westerman et al. (2012) sheds light on the relationship be-tween digital maturity (inside the DI and TMI framework) and financial performance.184 publicly traded companies were analyzed in order to establish a connection be-tween digital maturity and financial performance.

In short, they find that firms, which are either high DI or high TMI, are outper-forming firms which are neither high DI nor high TMI. Subsequently, firms catego-rized as Digirati have the highest performance, leaving not only Beginners behind butalso firms that have only either high DI or high TMI. The authors compare financialfigures such as revenue generation, profitability, and market valuation.

They find that firms with either high DI, high TMI or both, derive more revenuefrom physical assets measured by fixed asset turnover and revenue/ employee multiple.Figure 2 illustrates the individual digital maturity levels for firms in the study withinthe DI and TMI framework. Digirati have, on average, a nine percent higher revenue

Figure 2: Scatter plot of 184 publicly traded firms and their corresponding digital maturity levels.The comparison value is the revenue generating abilities of these firms. High digital intensity isan indicator that firms generate revenue more efficiently (Westerman et al., 2012, p. 7.).

generation compared to the average firm. The researchers argue that these firms havethe ability to manage more volume with the same physical capacity. It appears thatmore advanced technologies, such as tracking demand in real time, help to utilizephysical assets more optimally. The allocation of human and physical assets is also

28

improved. When it comes to profitability, firms with low TMI, regardless of DI levelsare worse off compared to firms that have high TMI levels. Therefore, firms withhigher TMI are more profitable. Digirati and Digital Conservatives are, on average,26 and 9 percent, respectively, more profitable compared to the average firm. Theresearchers argue that a clear vision of the new digital future aligns investments.Moreover, firms with high TMI levels have a higher market valuation (measured asPtB ratio) compared to firms that do not. In general, firms that are more matureon either the DI or TMI dimension are outperforming (Westerman et al., 2011). Inshort, there is a pronounced gap between digital leaders and organizations that lackdigital maturity, which leads to a growing opportunity cost for the latter. The resultsof the study are clear, however, they must be treated with some reservation since notevery industry is affected equally.

In a study conducted by Bughin et al. (2017), firms’ digital intelligence is mappedon a 100-point scale. Subsequently, they categorize firms into three groups with regardto their digital intelligence scores, namely, digital leaders (score of 41 and above), av-erage firms (score between 25 and 40) and laggards (score below 25). In their study,the average firm achieves a score of 34. Their findings are as follows. First, firms inthe top group, the digital leaders, are capable of repelling the digital pressure andusing digitalization to their advantage. Their scores are highly correlated with margingrowth and revenue. Second, firms labeled as average do neither profit nor experiencenegative growth. This group also represents the majority of companies with around 60percent. Bughin et al. (2017) state that these companies neither benefit from digital-ization, nor are they affected by the digital disruption. Third, laggards underperformand are faced with shrinking revenue and negative growth profiles (see also Fitzgeraldet al., 2013). Moreover, Weill and Woerner (2015), in a recent study for the MIT Cen-ter for Information Systems Research, find that firms with a profound engagement indigital ecosystems outperform their industry peers. Numerically, companies with atleast 50 percent of their revenue from a digital ecosystem, outperform competitors by32 percent in revenue growth and by a 27 percent increase in profit margins.29

Digital transformation is not affecting all industries at the same pace. On the onehand, some industries were hit early due to the arising digital competition. The musicindustry, for instance, adopted early to new digital concepts which emerged as new

29Similarly, in a study of the announcements of IT investments, Dehning et al. (2003) find positive,abnormal returns related to said announcements. They continue and argue that new IT investmentsare more likely than not able to produce out of the ordinary returns.

29

threats to traditional business strategies. On the other hand, some industries have yetto be affected by severe digital transformation. An example would be manufacturingwhich is an industry that traditionally reacts slowly to new changes. Other industriessuch as the insurance or retail sectors are somewhat in the middle. Generally speaking,industries that have a tight regulatory environment, risk-averse culture and complexorganizational structures, suffer limitations with regard to advanced technologicalinnovations (Westerman et al., 2011).

In conclusion, the general understanding of digitalization’s effect on firm perfor-mance is threefold. First, higher levels of digitalization normally benefit companieswhen it comes to profitability and imposes a competitive edge across the board. Sec-ond, the speed of digitalization is uneven and thus, companies that are ahead of theirpeers, reap disproportionate benefits. Third, for certain sectors, there is a develop-ment towards a winner-take-all dynamic which is a severe threat to companies thatfall behind (Fitzgerald et al., 2013; Bughin et al., 2017).

30

3 | Methodology and Data

This chapter aims to present the tools used in order to qualitatively explorethe effects of digitalization on firm performance. To begin with, this chapter gives adistinct overview of the composition of the digital maturity proxy. Next, it presents thesummary statistics for the firm data used. Additionally, it illustrates the empiricalmodels that are deployed for the empirical testing process. The model descriptionsection itself is split into two stages. The first part introduces the portfolio analysiswhich looks into the effects of digitalization on stock performance. The second partpresents the regression analysis approach which aims to answer the question, whetherdigital maturity has a profound impact on subsequent operating performance.

3.1 The Digital Maturity Measure

In my attempt to investigate the effect of different levels of digitalization, that isdigital maturity, on firm performance, a proxy needs to be found to adequately modeldigital maturity. Since digital maturity is not a single metric that simply can belooked up, such as the unemployment rate or the number of patents granted for acompany, it must be constructed manually. However, there is no real consensus asto what digitalization and digital maturity encompass. Moreover, information abouta company can be gathered either by an outside view or through a survey, e.g. aquestionnaire.

For my analysis, I use a measure created by von Blixen-Finecke et al. (2017) forthe digital maturity levels. In their study, von Blixen-Finecke et al. (2017) measuredegrees of digital maturity in six dimensions, namely, digital marketing, digital prod-uct experience, electronic commerce (E-commerce), electronic customer relationshipmanagement (E-CRM), mobile and social media.

31

In more detail, digital marketing assesses a firm’s ability to utilize search en-gine marketing and advertising in order to reach customers. Digital product expe-rience evaluates a firm’s online presence which includes functionality and content.E-commerce captures a firm’s capability to sell products via digital channels and ad-ditionally looks into the purchasing processes. E-CRM encompasses a firm’s skill toenhance customer relationships within digital channels, such as unique personaliza-tion for customers and customer engagement. The mobile category assesses a firm’sability to utilize mobile channels, which are mobile sites and mobile applications. So-cial media, finally, evaluates a firm’s engagement on social media platforms, such asFacebook or Twitter (von Blixen-Finecke et al., 2017).

According to MIT Center for Digital Business, the key areas transformed as resultof digital innovations in organizations are the following: business model, operationalprocess and customer experience (Westerman et al., 2011). The first building block,the business model, includes product augmentation, transitioning from physical todigital, and new digital business such as digital products. These digital modificationselevate the organizational boundaries and pave the way for digital globalization withshared digital services. Digital products are often supplementing the previously offeredproducts by a company. Going digital also means offering mobile apps as a substitutefor a company’s internet website (Tiwana et al., 2010; Westerman et al., 2011).

The second building block encompasses operational processes which include pro-cess digitalization such as new analytics that improve performance, knowledge sharingand faster communication, and an increase in operational transparency. In short, theperformance management is streamlined. With new digital innovations, such as self-service systems, automation does not stop at production but can be extended to tasksand processes where human interaction was previously involved but is no longer re-quired. The digital transformation also affects performance management at its core.By measuring and storing an increased amount for data, performance transparencyand the level of detail shown by analytics are increasing. Managers can rely on realfigures often close to real-time and can compare key figures across sites. Powerful anal-ysis tools, which turn information stored in data into insights, can help organizationsto gain a strategic advantage (Westerman et al., 2011; Weill and Woerner, 2018).

The last building block is customer experience. Digital transformation is capableof enhancing and deepening the experience for customers. Whereas transformed op-erational processes are mostly internal processes that are invisible for third parties,

32

changing the customer experience is more visible. For organizations, it is important tobetter understand their customers. Customer demographics can be analyzed in-depthin order to improve customer processes, marketing, and digital sales. Organizationscan launch social media campaigns to promote a new product and to target a veryspecific audience. Moreover, touch points with customers, such as customer services,need to be updated. This might include building an online community where cus-tomers stay connected with each other and share their experience with the product.30

Creating touch points with customers strengthens and extends the brand. Moreover,organizations invest heavily in analytics in order to gain additional knowledge abouttheir customers. Through customer analytics, companies can offer customized prod-ucts and personalize customer service (Westerman et al., 2011).

All in all, the measure for digital maturity is constructed as follows (von Blixen-Finecke et al., 2017).

DMi = 0, 2DIMi + 0, 2DPEi + 0, 2ECi + 0, 15CRMi + 0, 15MOi + 0, 1SMi, (1)

where DMi is the final value for digital maturity for firm i and where DIMi is thevalue for digital marketing, DPEi is the value for digital product experience, ECi

is the value for e-commerce, CRMi is the value for e-customer relationship manage-ment, MOi is the value for mobile and SMi is the value for social media for firm i,respectively.

I can be fairly confident that with the key areas outlined by the MIT Center forDigital Business and the measures put forth by von Blixen-Finecke et al. (2017), Iam able to classify companies according to their digital maturity adequately. Theweighting is to some extent discretionary, but overall it should reflect digital matu-rity adequately as it touches upon at least two of the three major building blockssuggested by the MIT Center for Digital Business for digitalization. Admittedly, theproxy rather heavily reflects companies relationships with customers and, arguably,neglects internal processes, such as big data analytics. However, I believe, given thecomplexity of digitalization and its impact on virtually all different business practices,it is suitable to use in my analysis.31 The digital maturity (DM) measure is evaluated

30Major car manufacturing companies, such as Mercedes Benz or BMW, have a very active pres-ence on Facebook where they engage with customers, potential customers and fans alike. The carmanufacturers regularly display pictures of their cars taken and shared by owners.

31To my knowledge, there are no metrics available that capture all facets of the digital transfor-mation. Especially the transformation of internal processes of a company still remains a black box

33

and adjusted annually for each firm in the sample period (von Blixen-Finecke et al.,2017). Therefore, changes in the levels of digitalization over time are incorporated.Figure 3 shows the distribution of the individual firms’ average digital maturity valuesin the sample. There are substantial differences in the digital maturity levels, whichis convenient for the ranking and exploitation for the analysis.

Figure 3: Overview of the average individual digital maturity values for each firm in the sample.

87.576.565.5

2

4

6

8

Digital maturity intensity (on a scale from 4 to 10)

Num

ber

offir

ms

The values for DM go as low as 5.5 for the least digital mature firms and as highas 8.0 for the digital leaders. The DM values for most firms are between 6.5 and 7.0.The median is 6.5 and firms categorized as laggards have an average DM value of5.6. On the other hand, the digital leaders are firms with an average DM of 7.3.32

Overall and with regard to the scaling from four to ten, it is obvious that the sampledistribution of firm digitalization levels are slightly skewed to the left.

for most outsiders.32An in-depth categorization of firms according to their DM rank follows in the Model Description

section (see chapter 3.3).

34

3.2 Summary Statistics

I gather publicly available yearly firm data and monthly stock returns for Swedishcompanies from Thomson Reuters Datastream. The measure for DM is taken fromvon Blixen-Finecke et al. (2017) and Statista. Table 1 reports the summary statisticsfor the firms in the sample. The sample consists of 24 Swedish firms which are listedon the NASDAQ OMX Nordic (both large and mid cap). The time period is from

Table 1: Sample data description.

V ariable Median SD

Return on Assets 5.04 5.57Cash flow to Sales 11.2 14.77Digital maturity score 6.50 0.72Total assets 7 566 180 242Capital Expenditure 258 441Market value 7 026 12 966Operating Profit Margin 11.37 9.91The sample period is 2010-2016.

Data source: Datastream. Return on assets, cash flow to sales,and operating profit margin are reported in percent. Totalassets, CapEx and market value are reported in MUSD.

the beginning of 2010 until the end of 2016, totaling 168 observations.33 The firmsare predominantly from the following sectors: communication, consumer, financialservices, industrial, retail, technology, and autos.

Table 2 shows the Spearman and Pearson correlation coefficients between digitalmaturity and various firm characteristics of interest in the analysis, such as ROA orln(TA). Strikingly, the association between the DM measure and ROA is slightlynegative. In detail, the correlation’s magnitude is negative 0.172 and negative 0.119,for the Spearman correlation and the Pearson correlation, respectively. The corre-lation, however, is only significant for the Spearman correlation coefficient and the

33The data for stock returns is differing (time period from January 2013 to December 2016).

35

Table 2: Pearson and Spearman correlation coefficients of DM and firm characteristics.

Spearman (below-diagonal) and Pearson (above-diagonal) correlation coefficientsDM ROA ∆ROA CFS ln(TA) ln(MV )

DM -0.119 -0.115 -0.188φ -0.142φ -0.050ROA -0.172φ 0.396φ 0.034 0.091 0.221φ

∆ROA 0.017 0.114 0.071 0.011 -0.070CFS -0.276φ 0.124 0.169φ 0.231φ 0.118ln(TA) -0.238φ 0.183φ -0.092 0.270φ 0.262φ

ln(MV ) -0.229φ 0.101 -0.078 0.283φ 0.882φ

This table reports the correlation coefficients for following variables. DM denotes thefirm’s digital maturity value, ROA the return of assets at time t, ∆ROA the change inreturn of assets from t−1 to t, CFS denotes cash flow to sales, ln(TA) is the logarithm ofthe firm’s total assets and ln(MV ) is the logarithm of the firm’s market value. φ denotescorrelation coefficients that are significant at the 0.1 level.

magnitude is far from being clear without ambiguity. Furthermore, the correlationcoefficients between operating performance, or ROA, and firm size measured in totalassets ln(TA) and market value ln(MV ) are positive for both the Spearman andPearson correlation. As expected, the association between total assets and marketvalue is positive and of a high magnitude, especially for the Spearman correlationcoefficient with 0.882.

3.3 Model Description

3.3.1 Portfolio Analysis

I construct six portfolios in order to analyze whether or not digitally mature compa-nies are outperforming peers in the stock market.

I construct the portfolios as follows.

µP F =n∑

i=1xi µi (2)

where µP F is the return for the portfolio PF , µi is the return of security i, xi isthe portfolio weight of security i in portfolio PF . The portfolio return is, thus, the

36

weighted average security return. The weighting is based on the market capitalizationin year t, which yields two size groups34, the small (S) and big (B) portfolios. In thenext step, I sort the companies according to their DM rank and categorize each intoone of three groups, based on the 30th and 70th percentiles of the corresponding DM

rank. Based on their DM rank, these portfolios subsequently are labeled low (L),middle (M) and high (H). Therefore, six size-DM portfolios are formed through theintersections: BH, BM , BL, SH, SM and SL. Following Hirshleifer et al. (2013),I hold these portfolios over a period of twelve consecutive months and compute themonthly size-adjusted returns of the low, middle and high DM portfolios using thefollowing formulas.

(S/L + B/L)2

(3)

(S/M + B/M)2

(4)

(S/H + B/H)2

(5)

where S, B stand for small and big portfolios, respectively, and L, M and H indicatelow, middle and high levels of digital maturity, respectively. In order to calculatethe monthly excess returns35 of the size-adjusted portfolio returns, I use the Swedishone-month treasury bill rate for the corresponding months.36

Additionally, I use the Sharpe (1966) ratio in order to analyze whether there aresignificant differences in risk-adjusted portfolio performance between different levels ofdigital maturity. This ratio measures the average excess returns of a portfolio relativeto its volatility. This is naturally interesting since the ratio answers the questionwhether the risk-return relationship is adequate (Cochrane, 2000).

The Sharpe (1966) ratio is defined as follows.

SRP F = E[RP F ] − Rf

σP F

(6)

34Contrarily to Hirshleifer et al. (2013) who sort firms into size groups based on the NYSE medianbreakpoint, I have to adjust the size breakpoint to the sample differently otherwise, and due to thelimited sample size, there would have been a disbalance in favor for large firms.

35Excess does not connote that the returns are bound to be positive.36I retrieve the Swedish one-month treasury bill rates directly from the official website of the

Swedish central bank (Riksbank, 2018).

37

where SRP F is the Sharpe ratio for portfolio PF , E[RP F ] is the average re-turn of portfolio PF , Rf is the risk free rate and σP F is the standard deviationof portfolio PF . In order to ascertain that the Sharpe ratio results are robust andmeaningful for a comparison, I determine the significance of the ratios. Hence, I deploya significance test suggested by Jobson and Korkie (1981) and amended by Memmel(2003). The hypothesis for testing for equality is as follows.

H0 = SRi − SRj = 0 (7)

where the null hypothesis H0 states that the Sharpe ratios SRi and SRj are equaland thus, do not differ from each other, statistically speaking.

For the hypothesis test, I compute the following z test statistic.37

z = SRij√ν

∼ N(0, 1) (8)

3.3.2 Risk Factor Models

Moreover, I use the Fama and French (1993) three-factor model and the Carhart(1997) four-factor model to analyze whether the size-adjusted returns of the DM

portfolios are captured by risk factor models.The Fama and French (1993) three-factor model looks as follows.

Ri,t − Rf,t = αi,t + βi,m(Rm,t − Rf,t) + βi,smbSMBt + βi,hmlHMLt + ηi,t (9)

where Ri,t is the return of portfolio i at time t, Rf,t is the risk free rate at time t,Rm,t is the return of the market portfolio at time t38, SMBt is the small minus bigfactor (small minus large firms) at time t and HMLt is the high minus low factor(high book-to-market minus low book-to-market stocks) at time t.

As for the second risk model, Carhart (1997) proposes the following four-factormodel which is an extension of the Fama and French (1993) three-factor model.39

37See Appendix A for more details about the significance test.38This factor, (Rm,t − Rf,t), is oftentimes labeled as MKT .39Findings by Jegadeesh and Titman (1993) suggest that past year ‘winners’ outperform compared

to past year ‘losers’ and this momentum factor should also be included in the risk model.

38

The Carhart (1997) four-factor model is specified as follows.

Ri,t − Rf,t =αi,t + βi,m(Rm,t − Rf,t) + βi,smbSMBt

+ βi,hmlHMLt + βi,momMOMt︸ ︷︷ ︸Added momentum factor

+ηi,t(10)

where Ri,t is the return of portfolio i at time t, Rf,t is the risk free rate at time t,Rm,t is the return of the market portfolio at time t, SMBt is the small minus big factor(small minus large firms) at time t, HMLt is the high minus low factor (high book-to-market minus low book-to-market stocks) at time t and MOMt is the momentumfactor (high minus low momentum return on portfolios) at time t. In this four-factormodel, the α is an estimate of net returns earned by the portfolio after adjusting forrisk.

3.3.3 Panel Data Regression

To examine the relation of digital maturity with operating performance more thor-oughly, I use cross-sectional time-series data and deploy dynamic panel data regres-sions.40 In general, a panel data regression can be distinguished from both, time-seriesor cross-section regressions, by a double subscript on its variables. Furthermore, dy-namic panel data models differ from standard panel data models by the presence ofa lagged dependent variable which indicates the dynamic relationship.

The general form of dynamic panel data models looks as follows (see Baltagi, 2005),in equation form:

Yi,t = γYi,t−1 + X1i,tβ1 + · · · + XKi,tβK + ηi,t with i = 1, . . . , N ;

t = 1, . . . , T(11)

where Yi,t is the dependent variable with entity i at time t,Yi,t−1 is the laggeddependent variable, Xi,t are the independent variables and ηi,t is the error term.

40According to Hsiao (2003), benefits from using panel data include, but are not limited to, theability to give more informative data, less collinearity and more efficiency and thus, more reliableestimates for parameters can be produced with more informative data. Naturally, the requirementsand assumptions for panel data regressions have to hold.

39

My approach to investigate the relationship between operating performance anddigitalization draws from Hirshleifer et al. (2013), where the relationship betweeninnovative efficiency and operating performance is investigated. In more detail, formy analysis I run the following specification, in equation form:

OPi,t+1 =β0 +Digital Maturity︷ ︸︸ ︷

β1DMi,t + β2ln

(1 + CapExi,t

TAi,t

)+ β3ln(MVi,t) + β4OPi,t

+ β5∆OPi,t + β6Technologyt + ηi,t

(12)

where OPi,t+1 is firm i’s operating performance in year t+1 measured in return onassets (ROA), DM is the digital maturity measure (see section 3.1), ln

(1 + CapEx

T A

)is the natural log of one plus capital expenditure to TA, ∆OPi,t is the yearly changein operating performance (year t and year t-1), ln(MVi,t) is the natural log of marketvalue of equity and Technology is a dummy variable that equals 1 for firms thatoperate in the technology, financial or communication sector and 0 otherwise.

Following Hirshleifer et al. (2013), I measure operating performance by ROA. Ad-ditionally, several controls are put in place. Similarly to Lev and Sougiannis (1996), Iinclude capital expenditure (CapEx) deflated by average total assets since a CapEx

ratio is found to explain operating performance to some extent41 (see also Panditet al., 2011). In addition, I use MV to proxy firm size account for competitive advan-tages (Kousenidis et al., 2000; Charitou et al., 2001). Moreover, the regression includestwo more controls that are current operating performance (OP ) and the change inoperating performance (∆OP ), which is the difference between current and previousOP . The theoretical motivation is as follows. Gu (2005) argues that operating perfor-mance has an inherit persistence, thus it should be used in the regression to capturethis effect (see also Pandit et al., 2011; Hirshleifer et al., 2013). Following Hirshleiferet al. (2013), I also add ∆OP to control for the change in operating performanceto capture the mean reversion in profitability (see also Fama and French, 2000).42

Lastly, the dummy variable is to control for sector effects.Because of some missing observations in my sample, I deal with a so-called unbal-

41Moreover, Lev and Sougiannis (1996) also use advertising expenditure ratios to explain firmoperating performance, however, I refrain from including it.

42Additionally, Bond (2002) states that, from an econometrical point of view, allowing for dynamicsin the regression process may be essential to receive consistent estimates of parameters, even though,the coefficients for the lagged dependent variable are not the main interest.

40

anced panel.43 Randomly missing observations in panels, however, should not affectthe outcome of a regression44 (Baltagi, 2005). Due to the lagged independent variable,I use the Generalized Method of Moments (GMM) estimation technique. Generallyspeaking, when it comes to dynamic panel data, GMM is considered to be superiorto other estimation techniques (Judson and Owen, 1999). More specifically, I use aGMM estimator proposed by Arellano and Bond (1991).45 This estimator handlesmodeling concerns well and prevents dynamic panel bias (Roodman, 2009).

43According to Baltagi (2005), missing observations in panel data models, unbalanced panels ingeneral, are quite common in economic empirical settings. He argues that unbalanced panels are,indeed, the norm in economic settings (Baltagi, 2005, p. 165).

44Ahrens and Pincus (1981) recommend an index to measure the unbalancedness, but this indexrarely ever surfaces and if, mostly in purely econometrical discussions.

45The Generalized Method of Moments was originally developed by Hansen (1982). For this GMMestimator, see also Arellano (2003).

41

4 | Empirical Research

This chapter presents the findings and results of the empirical testing.The first part illuminates the findings from a capital market point of view and issubdivided into two parts: portfolio testing and risk factor models. The second partpresents the results from the dynamic panel data regression and thus gives insightfrom an operating performance point of view.

4.1 Portfolio Analysis Results

4.1.1 Digital Maturity and Portfolio Returns

Table 3 presents the results of the portfolio analysis. In general, the portfolio analysisyields mixed results. Interestingly and contrarily to the proposed research question,more digitally mature firms’ returns are not visibly outperforming compared to theirpeers. In more detail, the monthly size-adjusted excess returns on DMlow portfoliosare 127 basis points. On DMmiddle portfolios the monthly size-adjusted excess returnsare 128 basis points. The monthly size-adjusted returns on DMhigh portfolios are 64basis points. The average digital maturity measures for DMlow, DMmiddle and DMhigh

portfolios are 5.59, 6.59 and 7.30, respectively. The entirety of the monthly averagesize-adjusted excess returns of the constructed portfolios are presented in table B1 (seeappendix B). Subsequently, DMhigh portfolios only outperformed low and middle DM

firms in one year (2014) during the sample period. During the entire sample period,firms with high DM values outperform firms with medium DM values and firmswith low DM values, 23 and 18 out of 48 months in total, respectively. Therefore,over the four year sample period, high DM firms did not outperform either mediumor low DM firms. However, the difference between excess returns of high DM firmscompared to medium DM firms is not statistically significant (p=0.43). The same

42

holds for the comparison of excess returns between high DM firms and low DM firmswhere there is no statistically significant difference between the two (p=0.44).

Table 3: Digital maturity and monthly average portfolio returns.

DMportfolio ER2013 ER2014 ER2015 ER2016 ERoverall

Excess returns in percentLow 1.43 0.80 1.62 1.25 1.27∅DM 5.54 5.67 5.70 5.70 5.59

Middle 1.57 0.93 0.90 1.70 1.28∅DM 6.57 6.72 6.70 6.38 6.59

High -0.13 1.30 0.16 1.23 0.64∅DM 7.24 7.40 7.41 7.13 7.30

This table presents the monthly average size-adjusted excess returnsof the constructed portfolios for the years 2013 to 2016. The portfolioconstruction is outlined in chapter 3.3.1. The excess return is calcu-lated as the difference between the Swedish Treasury bill rate andthe size-adjusted portfolio returns. Additionally, the average digitalmaturity (∅DM) values are reported for each portfolio and year.

In table 4 additional insight in form of the monthly average size-adjusted returns(excess returns), the standard deviation and the Sharpe ratio is displayed. The stan-dard deviation for DMhigh portfolios is close to being strictly smaller compared toDMmiddle portfolios and only falls short in one year compared to the DMlow portfolios.Firms with higher digital maturity seem to have less fluctuating returns. In detail, theaverage standard deviation for the DMhigh portfolio revolves around 3.48, whereas itis 4.45 for the DMlow portfolio. As for the Sharpe (1966) ratio comparison46, there isno distinct difference between the portfolios. According to this performance measure,the DMhigh portfolio does not outperform either the DMmiddle or DMlow portfolios.This means, although the volatility of the DMhigh portfolio is somewhat smaller com-pared to the DMlow portfolio, the smaller excess returns offset it.47 In short, theSharpe ratio results are ambiguous and the differences are most of the time not sta-

46Additionally, I use a closely related performance measure, the Modigliani and Modigliani (1997)’srisk-adjusted performance (RAP) criteria. The results can be found in table C1 in appendix C.

47The Sharpe ratio, as a risk-adjusted measure, will score higher, all else equal, underless volatility (see Opdyke, 2008).

43

Table 4: Results of portfolio analysis with returns, excess returns, stan-dard deviation and Sharpe (1966) ratios.

Year PF DMhigh PF DMmiddle PF DMlow

Average return2013 0.85 2.55 2.412014 1.76 1.39 1.262015 -0.14 0.61 1.322016 0.59 1.06 0.61Average excess return2013 -0.13 1.57 1.432014 1.30 0.93 0.802015 0.16 0.90 1.622016 1.23 1.70 1.25Standard deviation2013 3.52 3.90 3.872014 2.87 3.19 2.852015 3.70 5.66 6.062016 3.82 4.73 5.00

Results above are reported in percentSharpe ratio2013 -0.13 1.39 1.282014 1.56 1.01 0.972015 0.15 0.55 0.922016 1.12 1.24 0.86

This table shows the monthly average size-adjusted returns, monthlyaverage size-adjusted excess returns, the standard deviation and theSharpe ratios of the constructed portfolios.

tistically significant. Only the difference of the Sharpe ratios in the years 2013 (forboth DMhigh-DMlow and DMhigh-DMmiddle) and 2015 (only DMhigh-DMlow) are sta-tistically different from zero. In 2013, the difference is significant at the 0.01 level bothtimes. For 2015, the difference is significant at the 0.1 level. For all other years, infer-ence about performance cannot be made with statistical significance. The same holdswhen the entire period is investigated.48 Overall, it seems, although without sufficientstatistical significance, that digital maturity is not influencing stock performance.

48See tables A1 and A2 in Appendix A for all Sharpe ratio significance tests.

44

4.1.2 Digital Maturity and Risk Factor Models

The results from the Fama and French (1993) three-factor and the Carhart (1997)four-factor risk models with the DM portfolios are depicted in tables 5 and 6, respec-tively. The insights from the risk models are the following.

Table 5: Result of the DM portfolios and the Fama and French (1993)three-factor model.

Fama French three-factor modelDM Excess Return α MKT SMB HML

Low 1.27 0.99 0.71 -0.13 -0.69t (1.75) (4.43) (-0.37) (-2.70)

Middle 1.28 0.89 0.67 0.23 -0.03t (1.63) (4.38) (0.71) (-0.11)

High 0.64 0.55 0.54 -0.38 -0.03t (1.35) (4.72) (-1.54) (-0.14)

This table reports the monthly average size-adjusted excess returns tothe constructed portfolios and the risk factor loadings from regressingportfolio excess returns on factor returns. MKT, SMB and HML are themarket, size and book-to-market factors of Fama and French (1993). Ex-cess return and α are displayed in percent. Factor returns are obtainedfrom Kenneth French’s website.

The intercepts, or α are the active return of the portfolios given the portfolios’exposure to the risk factors. For the Fama and French (1997) three-factor model, theα is 0.99, 0.89 and 0.64 for the low, middle and high DM portfolios, respectively.This means all DM portfolios posted positive active returns, i.e. they outperformagainst the benchmark. As for the Carhart (1997) four-factor model, the α49 is 1.02,1.27 and 0.68 for the low, middle and high DM portfolios, respectively. Again, allDM portfolios perform above their corresponding risk levels and earn positive activereturns.50 In more detail, the DMmiddle portfolio’s α is slightly higher than the DMlow

and noticeably higher than the DMhigh and is significant at conventional levels. This49The α is commonly referred to as Carhart (1997)’s four-factcor alpha (Farah, 2002).50For the sake of brevity and to avoid tedious repetition of the factor models’ results, I will lay

the focus on the outcome of the Carhart (1997) four-factor model in the following paragraph. Thefactor loadings are very similar in both models in any case.

45

finding contradicts the statement that high digital maturity firms outperform firmswith lower digital maturity. However, the α values for DMlow and DMhigh portfolio arenot statistically significant, therefore, this interpretation stands on shaky ground. The

Table 6: Result of the DM portfolios and the Carhart (1997) four-factor model.

Carhart four-factor modelDM Excess Return α MKT SMB HML MOM

Low 1.27 1.02 0.71 -0.13 -0.70 -0.03t (1.62) (4.32) (-0.37) (-2.46) (-0.10)

Middle 1.28 1.27 0.64 0.23 -0.21 -0.40t (2.16) (4.18) (0.72) (-0.78) (-1.59)

High 0.64 0.68 0.53 -0.38 -0.09 -0.14t (1.53) (4.55) (-1.53) (-0.44) (-0.74)

This table reports the monthly average size-adjusted excess returns to the con-structed portfolios and the risk factor loadings from regressing portfolio excess re-turns on factor returns. MKT. SMB and HML are the market, size and book-to-market factors of Fama and French (1993). MOM is the momentum factor of Carhart(1997). Excess return and α are displayed in percent. Factor returns are obtainedfrom Kenneth French’s website.

risk factor loadings are as follows. First, the market risk factor, MKT (or RM,t −Rft)which shows the sensitivity of the DM portfolios to the market, is positive and smallerthan 1 for all DM portfolios. In detail, the MKT is 0.71, 0.64 and 0.53 for the low,middle and high DM portfolios, respectively. This means all portfolios move withthe market, but are, in general, less sensitive to movements (MKT < 1).51 Addi-tionally, the positive market factor strictly decreases from low to high, which meansPFhigh are less sensitive to the overall market compared to the other DM portfolios.In other words, the more digital a firm, the smaller the MKT effect. Admittedly,this difference in MKT factors is relatively small. Nonetheless, the relation is highlystatistically significant. Second, the size factor, SMB which shows the sensitivity ofthe DM portfolios to firm size effect, is small but varies between a negative relationfor DMlow and DMhigh portfolios and a positive relation for the DMmiddle portfolio.More specifically, the SMB is -0.13, 0.23 and -0.38 for the low, middle and high DM

51Generally speaking, equity-only funds tend to have MKT values around 1.

46

portfolios, respectively. Third, the value factor, HML which shows the sensitivity ofthe DM portfolios to the value effect, is negative for all DM portfolios. In detail,HML is -0.70, -0.21 and -0.09 for the low, middle and high DM portfolios, respec-tively. The value factor strictly increases from the DMlow portfolio to the DMhigh

portfolio where it is very close to zero. This could be interpreted that low DM firmsare more likely to be growth stocks whereas high DM firms are less likely to growand converge more closely to value stocks, while technically still being in the betweengrowth and stock categorization. In other words, with higher digital maturity, firmsleave the realm of growth stocks and approach value stocks. However, only the HML

factor for the DMlow portfolio is statistically significant at conventional levels, thus,this interpretation is to be treated very cautiously. Fourth and lastly, the momentumfactor, MOM which shows the relation between past returns and the DM portfolios,is negative for all DM portfolios. There is little difference between the returns for theDMlow and DMhigh portfolios according to the data. Moreover, the MOM factors arenot statistically significant.52

In conclusion, the data from portfolio testing and risk models does not supportthe hypothesis that digitally mature firms have superior stock performance comparedto less digitally mature ones. The results are inconclusive at best.

4.2 Panel Data Regression Results

This section presents the results from the dynamic panel data regression. A visualrepresentation of the panel data regression results with the average slopes, interceptand corresponding t-values is given in table 7 (see also table D1 in appendix D).Interestingly and inconsistent with most of the theoretical findings, the results indicatea significantly negative relation between the DM measure and future ROA. This effectis significant at the 0.05 level.53 This suggests that higher levels of digital maturity

52Additionally, I tried to implement the Carhart (1997) four-factor model augmented with theUMO factor (undervalued minus overvalued) introduced by Hirshleifer and Jiang (2010). The au-thors argue UMO factor loadings help identify stock misvaluation. This augmentation is suggestedby Hirshleifer et al. (2013) in order to analyze whether or not portfolios are under- or overvalued.Unfortunately, the data for the UMO factor is only provided until the year 2015 which reduces myalready small sample period by another year. Also, after running the regressions, the UMO factorsfor all three DM portfolios were indistinguishable from zero and of little significance.

53This finding affirms the negative correlation for both the Spearman and Pearson correlationcoefficient between digital maturity and operating performance. To recall, the coefficients are -0.172for the Spearman correlation and -0.119 for the Pearson correlation.

47

reduce subsequent operating performance slightly which is contrary to the findingsput forth by Westerman et al. (2012) and Bughin et al. (2017). The negative coefficientfor DM , however, ties into findings by Scott et al. (2017), where a temporary negativerelationship between DM and firm performance is hypothesized.

Consistent with literature, the highly significantly positive slope on ROA andthe highly significantly negative slope on ∆ROA confirm, in each case, persistenceas well as mean reversion in firm operating performance (Hirshleifer et al., 2013;Pandit et al., 2011; Gu, 2005). These effects are significant at the 0.01 level. Capital

Table 7: Digital maturity and subsequent operating perfor-mance panel regression results.

Operating PerformanceROA

Intercept 9.22(3.51)∗∗∗

DM -0.88(-2.01)∗∗

ln(CapEx/TA) -1.75(-0.18)

Current ROA 0.65(7.65)∗∗∗

∆ROA -0.39(-12.54)∗∗∗

ln(MV ) -0.13(-2.23)∗∗

Technology -1.16(-2.79)∗∗∗

Adjusted R2 0.48The panel data regression results are reported in this table.Heteroscedasticity robust t-statistics are reported in paren-theses. ∗∗∗, ∗∗ and ∗ indicate the statistical significancelevels at p < 0.01, p < 0.05 and p < 0.10, respectively.

expenditure correlates negatively with subsequent operating performance, however,the relation is not statistically significant at conventional levels. Typically, capitalexpenditure is expected to benefit future profitability (Lev and Sougiannis, 1996). Asfor the controls for market value and firms, I find that the slope for natural log of

48

MV is very close to zero, yet significant at the 0.05 level. It appears that firm sizedoes not affect operating performance for the firms in the sample. In literature, thereis plenty of evidence for both negative and positive effects of firm size (Hansen andWernerfelt, 1989; Amato and Wilder, 1985).54 In addition, the slope for the dummyvariable is negative and significant at the 0.01 level. A slight underperformance isattributable to technology-heavy industry sectors.

In summary, the results of the dynamic panel data regression show that digital-ization is not positively related to subsequent firm operating performance. In fact, itis weakly negatively associated with the following year operating performance. Again,this in contrast to the suggested positive effect of digitalization on firm performance,i.e. the general concept of a digital advantage is not confirmed. I will further elaborateon the peculiarities and implications of the findings in the following chapter.

54The differences depend on the model set-up as well as the theories used to interpret the results.

49

5 | Discussion

This chapter further illustrates and analyzes the results from the em-pirical testing for both, the portfolio analysis and the regression model. First, aninterpretation of the results is presented. Second, findings from the theoretical back-ground are tied into the empirical testing results in order to corroborate the discussion.Ultimately, the goal of this chapter is to arrive at a comprehensible understanding ofthe findings. In addition, the limitations of the thesis are addressed.

5.1 Interpretation of the Results

To begin with, existing literature has emphasized the beneficial role of digitaliza-tion on overall firm performance. However, the ambiguous results from the empiricalresearch in this thesis impede a clear insight into the expected effects of digitaliza-tion on firms’ stock returns and operating performance. The data from the portfolioanalysis suggests that there is no clear relationship between digitalization and stockreturns. In addition, empirical factor pricing models, such as the Fama and French(1997) three-factor model and the Carhart (1997) four-factor model, do not entirelyexplain the DM-return relation. The result from the dynamic panel data regressionsuggests that digital maturity is weakly negatively associated with firm operatingperformance. These results, however, give rise to the question as to why there was nocoherent outcome.

More in-depth, my results do not substantiate the findings of various studies55

(Westerman et al., 2012; Bughin et al., 2017; Weill and Woerner, 2015). Because ofa different setup and focus of the studies, the findings are not directly comparable toeach other, yet, coherent inferences can still be made. Westerman et al. (2011) in their

55For the terminology see chapter 2.6.1.

50

study find that profitability and revenue generation are higher for firms with above av-erage DI values. Numerically, digital leader’s profitability is about ten percent highercompared to the average firm. Weill and Woerner (2015) report an increase of around30 percent in revenue growth and a 27 percent increase in profit margins for compa-nies that are embracing digital technology and operate within the digital ecosystem,compared to direct competitors. In contrast to these findings, I cannot observe a pat-tern or relationship between digital maturity and profitability. Somewhat similar tomy study set-up, Bughin et al. (2017) split firms into three groups according to theirdigitalization ranking. They find that digital leaders gain positive revenue growth,whereas average firms neither profit nor experience negative growth and laggardsface shrinking revenue. In short, they find a significant correlation between digitalleaders and revenue growth. Contrarily, in my results digitally mature firms do notoutperform their peers, when, in fact, DM is linked to a weaker subsequent operatingperformance.

A possible explanation is as follows. Some changes due to digital innovations andtechnology transformations will not be visible in the short run. Their effects seam-lessly merge into present business practices and thus make it increasingly difficult topinpoint their actual contribution to firm performance (Weill and Woerner, 2018).Moreover, information about digital maturity is difficult to process. In general, ICTadvances affect changes in firm structure positively (Spanos et al., 2002) and con-tribute to productivity and economic success of organizations (Hernaus et al., 2012).Realistically speaking, not all firms will benefit from digital innovations.

In their study about innovative efficiency and stock returns, Hirshleifer et al.(2013) find robust evidence that more innovative firms outperform peers. Needlessto say, digital maturity levels do not necessarily correspond to innovative efficiencylevels, however, my methods are related to their approach. Therefore, I believe someinference can be made. Whereas innovative efficiency (measured in patents or citationsand corresponding R&D levels) is generated internally, digital maturity can be raisedby implementing and using available external systems. It is one thing for a company touse, for instance, customer demographics in order to enhance marketing and digitalsales. But to actually reap the benefits from doing so is another thing. This mayeven be an intractable process for some firms. Weill and Woerner (2018) bluntlydescribe this predicament as firms trying to pursue digital transformation “withoutany sense where they are going” (p. 21). Thus, the positive outcome of digital maturity

51

is not only dependent on the implementation of new digital solutions but also onthe subsequent use internally. However, an in-depth study that quantifies firm stockperformance of digitalization effects has yet to be published.

The above-mentioned shortcoming of digital maturity and its effect on firm per-formance is one of the reasons why Westerman et al. (2012) use DI and TMI in theirframework. Firms with high DI and low TMI values would still be considered to bedigitally mature, however, they often miss the consequential step of implementingadequate governance internally. This dampens the outcome due to the suboptimalutilization of digital efforts. Nevertheless, in their study Westerman et al. (2012) findthat firms with high DI, regardless of their TMI, still perform at higher levels com-pared to firms with low DI, again regardless of their corresponding TMI. Anotherpossible explanation follows the findings of Scott et al. (2017). In their study, theauthors, indeed, find robust evidence that technology innovations have a positive im-pact on profitability, however, they also argue that these changes take time to unfoldtheir full potential. Scott et al. (2017) find robust evidence that an adoption of a newdigital innovation reduces long-term returns (measured in return on sales) for up tofour years, before the positive effects prevail. In conclusion, the long run effect of dig-ital innovations on profit margins is positive after the initial drawbacks are overcome.Changing internal processes is no easy task and especially the fine-tuning as well assynchronizing are time-consuming. Additionally, replacing processes at the core of abusiness can be a multiyear undertaking (Weill and Woerner, 2018). In any case, somechanges are inevitable and firms that do not transform accordingly will presumablybe put out of business, or suffer from reduced growth perspectives (Fitzgerald et al.,2013).

5.2 Limitations, Proxy and Sample Implications

My findings are subject to several limitations that need to be taken into considerationfor the interpretation of the results.

My digital maturity factors condensed into the digital maturity proxy might notcapture all the details accurately enough and therefore digital maturity might not beadequately captured. This has implications for firm sorting for the portfolio construc-tion as well as individual firm performance. However, the paucity of digital maturitymeasures owing to the cumbersome collection process of meaningful variables makes

52

it difficult to obtain an impeccable measure. In addition, the digitalization of firmsis not a uniform transformation which makes it difficult to access different levels ofdigital maturity with great accuracy. Thus, finding an adequate proxy is virtually aherculean task.56 Even though I tried to triangulate the digital maturity proxy withvarious papers and it seems to be adequately well defined for the analysis, it is unclearhow big the impact of additional factors, such as big data analytics or managerialaspects, would have been.57 Including different proxies, or with altered emphasis, mayseem innocuous at first, but the results could differ and the conclusions drawn wouldnot be the same. Strictly speaking, dissimilar results with a different proxy cannot beruled out. However, this data can only be retrieved via surveys and extensive ques-tionnaires which would have been beyond the scope of this thesis. As for the measurefor firm performance itself, although return on assets is widely regarded as a keyperformance metric, no single metric is faultless. Thus, another performance measuresuch as cash flows would not be amiss.

Furthermore, the sample I used in this thesis is relatively small (due to the limitedaccessibility of digital maturity measures) and this could have implications for theresults. A limited sample size is by definition prone to extreme values and individualfirms that are for some unspecific reason under- or overperforming might distort thedata. Despite my efforts to find a suitable sample for testing, it is a possibility thatthe results are to some extent driven by individual firm performance independentof digital maturity. Generally speaking, the testing would positively benefit from ahigher number of observations.

Additionally, it is worth mentioning that the geographical scope and the focus onSwedish firms are acknowledged as another limitation. Traditionally, Swedish com-panies are on the forefront of technological change and new trends. Thus, it is verylikely that Swedish firms might be more digitally mature compared to the averageEuropean or US firm and the nuances between firms might be marginal.

56It is worth mentioning that there are reliable sources and accurate data collections of digital-ization levels. However, and especially because this information is of interest for a great many ofplayers, this information, more often than not, is not accessible for outsiders.

57As a short marginalia, since non-financial indicators are not subject to well-defined standards ofmeasurement, the relation between operating performance and non-financial indicators is not alwaysstraightforward (Gu, 2005).

53

Everything changes and nothing stands still.

Heraclitus, as cited in Plato’s Cratylus

6 | Conclusion

The primary goal of this thesis was to critically examine the effects of digi-talization on firm performance, thus, to advance the understanding of this topic. Itsmain purpose was to find a concise, yet thoroughly data-driven, answer to the crucialquestion whether or not a digital advantage exists.

By collecting data from various Swedish firms in different sectors, my researchsought to shed light on the effect of digital maturity on subsequent operating perfor-mance as well as stock returns. In an effort to analyze this relationship, I deployedtwo empirical methods. Namely, a portfolio analysis where firms of different sizes anddigital maturity levels were sorted into different portfolios and examined with regardto their stock performance, and a regression model which incorporated accountingmeasures and digital maturity levels. Furthermore, I examined various articles andpapers during an extensive and systematic investigation of research articles in orderto create a solid theoretical foundation for my empirical analysis. Generally speaking,there is very little academic work on digitalization’s effect on performance on firmlevel. The majority of studies revolve around a simple comparison of accounting mea-sures. Hence, the inherent objective of this research was to investigate the relationbetween digital maturity and firm performance empirically. My findings, however, donot support the hypothesis that digitally mature firms outperform their less digitallymature peers, i.e. I cannot corroborate the argument of a digital advantage.

Undoubtedly, the effects of digitalization are transforming firms and doing busi-ness will be subject to change in digital technologies. The digital revolution is inprogress, yet it is not foreseeable how far it will go. New innovative concepts, such asmachine learning, digital tokens58 or artificial intelligence are not limited to certain

58Digital tokens which are closely related to cryptocurrencies, can store information which subse-quently can be used for communication or payment; a step closer to, for instance, smart factories.

54

firms or sectors. They could potentially benefit all business entities to some extent.Unlike previous technological improvements, digital innovations are not bound to afirm’s total assets, access to raw materials or resources. Recent developments seemto further justify the beneficial aspects of digitalization (Loonam et al., 2018). Animportant consideration to keep in mind when looking at the findings in retrospect,is the complexity aspect of digitalization. The inherent expansive nature of digitalinnovations with their endless possibilities makes it difficult to arrive at a commonconsensus as to how digitalization should be quantified. Therefore, reaching a consen-sus on how to define and measure digitalization could be a wellspring for continuousdebate. This would also be advantageous for future research by making results morecomparable.

This thesis has shown that digitally mature firms’ stock returns do not differ sig-nificantly compared to returns of less digitally mature firms. Also, the results from thedynamic panel data regression indicate that digital maturity is linked to a lower subse-quent operating performance, therefore, digitalization’s conjecturable positive effectson firms are not reinforced. Further research in the area of digitalization should bemore exhaustive, in-depth and over an extended period of time, in order to substan-tiate the claims put forward by literature.

•

In concluding, my intent with this thesis was also to venture into uncharted ter-ritory and to add a small contribution to the research landscape. Even though thisthesis cannot provide a definite answer to the effects of digitalization on firm perfor-mance, it paves the way to arriving at a better understanding of one, if not the mostinteresting transformations for businesses and organizations in the years to come.

55

References

Ahrens, H. and Pincus, R. (1981). On Two Measures of Unbalancedness in a One-wayModel and Their Relation to Efficiency. Biometric Journal, 23:227–235.

Amato, L. and Wilder, R. P. (1985). The Effects of Firm Size on Profit Rates in U.S. Manufacturing. Southern Economic Journal, 52(1):181–190.

Anderson, M. C., Banker, R. D., and Ravindran, S. (2006). Value implications ofinvestments in information technology. Management Science, 52:1359–1376. Avail-able at https://doi.org/10.1287/mnsc.1060.0542.

Arellano, M. (2003). Panel Data Econometrics. Oxford University Press, Oxford.

Arellano, M. and Bond, S. (1991). Some tests of specification for panel data: MonteCarlo evidence and an application to employment equations. Review of EconomicStudies, 58:277–297.

Baltagi, B. H. (2005). Econometric analysis of panel data. John Wiley & Sons Ltd,England, 3:135–181.

Bank for International Settlements (2002). Statistics on payment and settlementsystems in selected countries. Available at https://www.bis.org/cpmi/publ/d49.pdf.

Barrett, M. and Walsham, G. (1999). Electronic trading and work transformation inthe london insurance market. Information Systems Research, 10:1–22.

Bharadwaj, A., El-Sawy, O., Pavlou, P. A., and Venkatraman, N. (2013). DigitalBusiness Strategy: Toward A Next Generation Of Insights. Management Informa-tion Systems Quarterly, 37(2):471–482.

Bond, S. (2002). Dynamic Panel Data Models: A Guide to Micro Data Methods andPractice. cenmap working paper CWP09/02, pages 1–27.

Brynjolfsson, E. and McAfee, A. (2014). The Second Machine Age: Work, Progressand Prosperity in a Time of Brilliant Technologies. W. W. Norton & Company,New York.

56

Bughin, J., Catlin, T., Hall, B., and van Zeebroeck, N. (2017). Improving your digitalintelligence. MIT SLoan Management Review, Digital. Available at http://mitsmr.com/2yCjHGM.

Bughin, J. and van Zeebroeck, N. (2017). The best response to digital disruption. MITSLoan Management Review, 58:80–86. Available at http://mitsmr.com/2oFlYgE.

Callewaert, P., Robinson, P. A., and Blatman, P. (2009). Cloud computing Forecastingchange. Deloitte.

Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. Journal ofFinance, 52(1):57–82.

Charitou, A., Clubb, C., and Andreou, A. (2001). The Effect of Earnings Permanence,Growth, and Firm Size on the Usefulnes of Cash Flows and Earnings in ExplainingSecurity Returns: Empirical Evidence for the UK. Journal of Business Finance &Accounting, 28(5):563–594.

Chen, L., Mislove, A., and Wilson, C. (2016). An Empirical Analysis of AlgorithmicPricing on Amazon Marketplace. International World Wide Web Conference Com-mittee. Available at http://dx.doi.org/10.1145/2872427.2883089 (Accessed: 2018-05-17).

Cloodt, M., Hagedoorn, J., and van Kranenburg, H. (2006). Mergers and acquisitions:Their effect on the innovative performance of companies in high-tech industries.Research Policy, 35:642–654.

Cochrane, J. H. (2000). Asset Pricing. Princeton University Press.

Comin, D. and Philippon, T. (2005). The Rise in Firm-Level Volatility: Causes andConsequences. NBER Macroeconomics Annual, 20:167–201. Available at https://doi.org/10.1086/ma.20.3585419 (Accessed: 2017-12-05).

Cukier, K. and Mayer-Schönberger, V. (2014). Big Data: A Revolution That WillTransform How We Live, Work, and Think. Eamon Dolan/Mariner Books.

Damodaran, A. (2014). 450 million users is the magic number of WhatsApp valu-ation. The Economic Times. Available at http://people.stern.nyu.edu/wgreene/entertainmentandmedia/WhatsApp-valuation.pdf (Accessed: 2018-05-17).

Davenport, T. H. (2014). Big data at work. Harvard University Press, pages 10–85.

Dehning, B., Richardson, V. J., and Zmud, R. W. (2003). The Value Relevance ofAnnouncements of Transformational Information Technology Investments. MISQuarterly, 27(4):637–656.

57

Desmet, D., Duncan, E., Scanlan, J., and Singer, M. (2015). Six Building Blocks forcreating a high-performing digital enterprise. McKinsey & Company, September.

Dobbs, R., Koller, T., Ramaswamy, S., Woetzel, J., Manyika, J., Krishnan, R., andAndreula, N. (2015). Playing to Win: The New Global Competition for CorporateProfits. McKinsey Global Institute, September:2–85.

Downes, L. and Nunes, P. (2014). Big Bang Disruption: Strategy in the Age ofDevastating Innovation. pages 16–18. Penguin Group.

Downes, L. and Nunes, P. F. (2013). Big Bang Disruption. Harvard Business Review,91(3):44–56.

Dutta, A., Lee, H., and Yasai-Arkedani, M. (2014). Digital systems and competitiveresponsiveness: The dynamics of it business value. Information and Management,51:762–773.

Estellés-Arolas, E. and de Guevara, F. G.-L. (2012). Towards an integrated crowd-sourcing definition. Journal of Information Science, 38:189–200.

Evans, D. S., Hagiu, A., and Schmalensee, R. (2006). Invisible Engines: How SoftwarePlatforms Drive Innovation and Transform Industries. The MIT Press, Cambridge,Massachusetts.

Fama, E. F. and French, K. R. (1993). Common Risk Factors in the Returns OnStocks And Bonds. Journal of Financial Economics, 33:3–56.

Fama, E. F. and French, K. R. (1997). Industry costs of equity. The Journal of Finan-cial Economics, 43:153–193. Available at https://doi.org/10.1016/S0304-405X(96)00896-3 (Accessed: 2017-12-05).

Fama, E. F. and French, K. R. (2000). Forecasting Profitability and Earnings. TheJournal of Business, 73:161–175.

Farah, N. (2002). The financial economics of performance measurement. PerformanceMeasurement in Finance, 1st edition:1–49.

Fitzgerald, M., Kruschwitz, N., Bonnet, D., and Welche, M. (2013). Embracing DigitalTechnology: A New Strategic Imperative. MIT Sloan Management Review, 55:1–12.

Focacci, A. (2017). Controversial curves of the economy: An up-to-date investigationof long waves. Technological Forecasting & Social Change, 116:271–285.

Fuentelsaz, L., Gómez, J., and Palomas, S. (2009). The effects of new technologies onproductivity: An intrafirm diffusion-based assessment. Research Policy, 38:1172–1180.

58

Gassmann, O., Frankenberger, K., and Csik, M. (2014). The St.Gallen Business ModelNavigator. Working Paper University of St.Gallen, pages 5–12.

Gimpel, H. and Röglinger, M. (2015). Digital Transformation: Changes and Chances -Insights based on an Empirical Study. Fraunhofer Institute for Applied InformationTechnology FIT, pages 5–19.

Gordon, F., Deighton, J., Ullrich, M., and Marcu, S. (2013). Banking in a DigitalWorld. A.T. Kearny, pages 2–19.

Gu, F. (2005). Innovation, Future Earnings, and Market Efficiency. Journal of Ac-counting Auditing and Finance, 20:385–418. Available at http://journals.sagepub.com/doi/abs/10.1177/0148558X0502000405 (Accessed: 2017-12-05).

Hansen, G. S. and Wernerfelt, B. (1989). Determinants of firm performance: Therelative importance of economic and organizational factors. Strategic ManagementJournal, 10(5):399–411.

Hansen, L. (1982). Large Sample Properties of Generalised Method of Moment Esti-mators. Econometrica, 50:1029–1054.

Haverkort, B. R. and Zimmermann, A. (2017). Smart Industry: How ICT Will Changethe Game. IEEE Internet Computing, 21(1):8–10.

Hays, C. L. (2004). What wal-mart knows about customers’ habits. The New YorkTimes. Available at https://nyti.ms/2kltC9B (Accessed: 2017-10-28).

Hernaus, T., Bach, M. P., and Vuksic, V. B. (2012). Influence of strategic approachto BPM on financial and non-financial performance. Baltic Journal of Manage-ment, 7:376–396. Available at https://doi.org/10.1108/17465261211272148 (Ac-cessed: 2018-04-21).

Hildebrandt, B., Handelt, A., Firk, S., and Kolbe, L. M. (2015). Entering the DigitalEra - The Impact of Digital Technology-related M&As on Business Model Innova-tions of Automobile OEMs. Research Paper, Thirty Six International Conferenceon Information Systems, Forth Worth.

Hirshleifer, D. and Jiang, D. (2010). A Financing-Based Misvaluation Factor and theCross-Section of Expected Returns. The Review of Financial Studies, 23(9):3401–3436.

Hirshleifer, D. A., Hsu, P.-H., and Li, D. (2013). Innovative efficiency and stockreturns. Journal of Financial Economics, 107:1–60. Available at https://ssrn.com/abstract=1799675.

59

Hodgson, G. M. (2014). What is capital? Economists and sociologists have changedits meaning: should it be changed back? The Cambridge Journal of Economics,38(5):1063–1086.

Hsiao, C. (2003). Analysis of Panel Data. Cambridge University Press, Cambridge.

Hylving, L., Henfridsson, O., and Selander, L. (2012). The Role of Dominant Design ina Product Developing Firm’s Digital Innovation. Journal of Information TechnologyTheory and Application, 13(2):5–21.

Jegadeesh, N. and Titman, S. (1993). Returns to Buying Winners and Selling Losers:Implications for Stock Market Efficiency. The Journal of Finance, 48(1):65–91.Available at https://doi.org/10.1111/j.1540-6261.1993.tb04702.x (Accessed: 2018-04-26).

Jobson, J. and Korkie, B. (1981). Performance Hypothesis Testing with the Sharpeand Treynor Measures. The Journal of Finance, 36(4):889–908.

Judson, R. A. and Owen, A. L. (1999). Estimating dynamic panel data models: aguide for macroeconomists. Economics Letters, 65:9–15.

Jun, S. (2008). The link between it investment and securities firms’ returns in korea.Journal of Economic Research, 13:1–43.

Kane, G. C., Palmer, D., Philips, A., Kiron, D., and Buckley, N. (2016). Aligning theorganization for its digital future. MIT Sloan Management Review, Summer:3–17.Available at http://sloanreview.mit.edu/digital2016.

Kane, G. C., Palmer, D., Phillips, A. N., and Kiron, D. (2015). Is your businessready for a digital future. MIT Sloan Management Review, 56:37–44. Available athttp://mitsmr.com/1BkggSi.

Kerin, R. A., Varadarajan, P. R., and Peterson, R. A. (1992). First-Mover Advan-tage: A Synthesis, Conceptual Framework, and Research Propositions. Journal ofMarketing, 56(October):33–52.

Kessler, A. M. (2015). Tesla adds high-speed autonomous driving to its bag of tricks.The New York Times. Available at https://nyti.ms/2oo15G4 (Accessed: 2017-12-05).

Korotayev, A., Zinkina, J., and Bogevolnov, J. (2011). Kondratieff waves in globalinvention activity (1900-2008). Technological Forecasting & Social Change, 78:1280–1284.

Kousenidis, D. V., Negakis, C. I., and Floropoulos, I. N. (2000). Size and book-to-market factors in the relationship between average stock returns and average

60

book returns: some evidence from an emerging market. The European AccountingReview, 9(2):225–243.

LaValle, S., Lesser, E., Shockley, R., Hopkins, M. S., and Kruschwitz, N. (2011).Big Data, Analytics and the Path From Insights to Value. MIT Sloan Man-agement Review, 52:21–31. Available at https://sloanreview.mit.edu/article/big-data-analytics-and-the-path-from-insights-to-value/ (Accessed: 2017-12-05).

Lev, B. and Sougiannis, T. (1996). The capitalization, amortization, and value-relevance of r&d. Journal of Accounting and Economics, 21:107–138. Availableat http://www.sciencedirect.com/science/article/pii/0165410195004106 (Accessed:2017-12-05).

Lieberman, M. B. and Montgomery, D. B. (1988). First-Mover Advantages. StrategicManagement Journal, 9:41–58.

Loebbecke, C. and Picot, A. (2015). Reflections on societal and business modeltransformation arising from digitization and big data analytics: A research agenda.Journal of Strategic Information Systems, 24:149–157. Available at http://dx.doi.org/10.1016/j.jsis.2015.08.002.

Loonam, J., Eaves, S., Kumar, V., and Parry, G. (2018). Towards digital transforma-tion: Lessons learned from traditional organizations. Strategic Change, 27(2):101–109. Available at https://doi.org/10.1002/jsc.2185 (Accessed: 2018-05-23).

Lucas, H. C., Agarwal, R., Clemons, E. K., El-Sawy, O. A., and Weber, B. (2013).Impactful Research on Transformational Information Technology: An OpportunityTo Inform New Audiences. MIS Quarterly, 37(2):371–382.

Luckerson, V. (2015). How Google Perfected the Silicon Valley Acquisition. Time.Available at http://time.com/3815612/silicon-valley-acquisition/ (Accessed: 2018-03-18).

Majchrzak, A. and Malhotra, A. (2013). Towards an information systems perspec-tive and research agenda on crowdsourcing for innovation. Journal of StrategicInformation Systems, 22:257–268.

Manyika, J., Ramaswamy, S., Khanna, S., Sarrazin, H., Pinkus, G., Sethupathy, G.,and Yaffe, A. (2015). Digital america: A tale of the haves and have-mores. McKinseyGlobal Institute, December:4–80.

McAfee, A. and Brynjolfsson, E. (2008). Investing in the it that makes a competitivedifference. Harvard Business Review, pages 99–107.

Memmel, C. (2003). Performance Hypothesis Testing with the Sharpe Ratio. FinanceLetters, 1:21–23.

61

MGI (2015). A Labor Market That Works: Connecting Talent with Op-portunity in the Digital Age. McKinsey Global Institute, June:2–85.Available at https://www.mckinsey.com/global-themes/employment-and-growth/connecting-talent-with-opportunity-in-the-digital-age (Accessed: 2017-12-05).

MIT Technology Review Custom (2016). The Rise of Data Capital. MIT TechnologyReview, pages 2–11. Available at http://files.technologyreview.com/whitepapers/MIT_Oracle+Report-The_Rise_of_Data_Capital.pdf (Accessed: 2017-12-05).

Modigliani, F. and Modigliani, L. (1997). Risk-adjusted performance. Journal ofPortfolio Management, Winter:45–54.

OED Online (2017). Digitization, n. Oxford University Press.

Oestreicher-Singer, G., Sundararajan, A., and Kane, G. C. (2017). The power of prod-uct recommendation networks. MIT Sloan Management Review, 59:7–10. Availableat http://mitsmr.com/2vvNzQe.

Opdyke, J. D. (2008). Comparing Sharpe Ratios: So Where are the P-Values? Journalof Asset Management, 8(5):308–336.

Pandit, S., Wasley, C. E., and Zach, T. (2011). The Effect of Research and De-velopment (R&D) Inputs and Outputs on the Relation between the Uncertaintyof Future Operating Performance and R&D Expenditures. Journal of AccountingAuditing and Finance, 26:121–144. Available at http://journals.sagepub.com/doi/abs/10.1177/0148558X11400583 (Accessed: 2017-12-05).

Pflaum, A. A. and Gölzer, P. (2018). The IoT and Digital Transformation: Towardthe Data-Driven Enterprise. IEEE Pervasive Computing, 17(1):87–91.

Picker, L. (2017). For Non-Tech Companies, if You Can’t Build It, Buy a Start-Up. The New York Times. Available at https://www.nytimes.com/2017/01/02/business/dealbook/mergers.html (Accessed: 2018-03-12).

Quah, D. (2003). Digital good and the new economy. CEP discussion paper, Centrefor Economic Performance, London School of Economics and Political Science,London:4–39. Available at http://eprints.lse.ac.uk/2236/ (Accessed: 2018-05-15).

Riksbank, S. (2018). Treasury bills. Available at https://www.riksbank.se/ (Accessed:2018-01-07).

Rogers, E. M. (1963). Diffusion of innovations. The Free Press of Glencoe, New York.

Roodman, D. (2009). A Note on the Theme of Too Many Instruments. Oxford Bulletinof Economics and Statistics, 71(1):135–158.

62

Sambamurthy, V., Bharadwaj, A., and Grover, V. (2003). Shaping Agility throughDigital Options: Reconceptualizing the Role of Information Technology in Contem-porary Firms. Management Information Systems Quarterly, 27(2):237–263.

Schoemaker, P. J. and Tetlock, P. E. (2017). Building a more intelligent enter-prise. MIT Sloan Management Review, 58:28–37. Available at http://mitsmr.com/2mkkzan.

Schumpeter, J. (1943). Capitalism, Socialism and Democracy. Routledge, Londonand New York.

Schwab, K. (2017). The Fourth Industrial Revolution. Crown Business.

Scott, S. V., van Reenen, J., and Zachariadis, M. (2017). The long-term effect of digitalinnovation on bank performance: An empirical study of swift adoption in financialservices. Research Policy, 46:984–1004. Available at http://www.sciencedirect.com/science/article/pii/S0048733317300616.

Shapiro, C. and Varian, H. R. (1999). Information Rules: A Strategic Guide to theNetwork Economy. Harvard Business School Press.

Sharpe, W. F. (1966). Mutual Fund Performance. The Journal of Business, 39(1):119–138.

Short, J. E. and Todd, S. (2017). What’s Your Data Worth? MIT Sloan ManagementReview, 58:17–19. Available at http://mitsmr.com/2mUheim.

Spanos, Y., Prastacos, G., and Pulymenakou, A. (2002). The relationship betweenInformation and Communication Technologies Adoption and Management. In-formation and Management, 39:659–675. Available at https://doi.org/10.1016/S0378-7206(01)00141-0 (Accessed: 2018-04-22).

Tiwana, A., Konsynski, B., and Bush, A. A. (2010). Platform Evolution: Coevolutionof Platform Architecture, Governance, and Environmental Dynamics. InformationSystems Research, 21(4):675–687.

Van Bommel, E., Edelman, D., and Ungerman, K. (2014). Digitizing the consumerdecision journey. Available at https://www.mckinsey.com/business-functions/marketing-and-sales/our-insights/digitizing-the-consumer-decision-journey (Ac-cessed: 2017-11-02).

Vanderbilt, T. (2013). The science behind the netflix algorithms that decide whatyou’ll watch next. WIRED Magazine.

Vogelsang, M. (2010). Digitalization in Open Economies. Springer-Verlag, BerlinHeidelberg.

63

von Blixen-Finecke, L., Byström, E. E., and Eriksson, A. (2017). Digital leaders inSweden 2018. BearingPoint Research, pages 4–59.

Weill, P. and Woerner, S. L. (2015). Thriving in an increasingly digital ecosys-tem. MIT Sloan Management Review, 56:27–34. Available at http://mitsmr.com/1BkdvAq.

Weill, P. and Woerner, S. L. (2018). Is Your Company Ready for a Digital Future?MIT Sloan Management Review, 59(2):21–25. Available at http://mitsmr.com/2A4yP1S (Accessed: 2018-05-21).

Westerman, G. (2017). Your Company Doesn’t Need a Digital Strategy. MITSLoan Management Review. Available at https://sloanreview.mit.edu/article/your-company-doesnt-need-a-digital-strategy/ (Accessed: 2017-12-05).

Westerman, G. and Bonnet, D. (2015). Revamping your business through digitaltransformation. MIT Sloan Management Review, Spring:2–5. Available at http://mitsmr.com/1vXayDu.

Westerman, G., Calméjane, C., Bonnet, D., Ferraris, P., and McAfee, A. (2011).Digital transformation: A roadmap for billion-dollar organisations. MIT Centerfor Digital Business and Capgemini Consulting, pages 5–62.

Westerman, G. and McAfee, A. (2012). The Digital Advantage: How Digital LeadersOutperform Their Peers in Every Industry. MIT Center for Digital Business.Available at http://digitalcommunity.mit.edu/docs/DOC-1105.

Westerman, G., Tannou, M., Bonnet, D., Ferraris, P., and McAfee, A. (2012). TheDigital Advantage: How digital leaders outperform their peers in every industry.MIT Center for Digital Business and Capgemini Consulting, pages 2–21.

White, L. (2002). Trends in Aggregate Concentration in the United States. Journalof Economic Perspectives, 16:137–160.

World Economic Forum (2016). World economic forum white paper dig-ital transformation of industries. White Paper, pages 6–27. Avail-able at http://reports.weforum.org/digital-transformation/wp-content/blogs.dir/94/mp/files/pages/files/digital-enterprise-narrative-final-january-2016.pdf.

Yoo, Y. (2010). Computing in everyday life: A call for research on experiential com-puting. MIS Quarterly, 34(2):213–231.

Yoo, Y., Boland, R. J., Lyytinen, K., and Majchrzak, A. (2012). Organizing for In-novation in the Digitized World. Organization Science, 23(5):1398–1408. Availableat https://doi.org/10.1287/orsc.1120.0771 (Accessed: 2018-03-15).

64

Yoo, Y., Henfridsson, O., and Lyytinen, K. (2010). The New Organizing Logic ofDigital Innovation: An Agenda for Information Systems Research. InformationSystems Research, December 2010:724–735. Available at https://doi.org/10.1287/isre.1100.0322 (Accessed: 2018-03-15).

Yu, L. (2006). Understanding Information Inequality: Making Sense of the Literatureand Digital Divides. Journal of Librarianship and Information Science, 38(4):229–252.

Zammuto, R. F., Griffith, T. L., Majchrzak, A., Dougherty, D. J., and Faraj, S. (2007).Information technology and the changing fabric of organization. Organization Sci-ence, 18:749–762.

65

Appendix A

Test for Significance (Sharpe Ratio)

This section presents the equations and background for the significance testing ofthe portfolio Sharpe ratios. The following hypothesis is used to test for the equalityof two Sharpe ratios:

H0 = SRi − SRj = 0 (13)

The test statistic is as follows:

z = SRij√ν

∼ N(0, 1) (14)

Jobson and Korkie (1981)’s variance of the statistic with Memmel (2003)’scorrection is given by (see also Opdyke, 2008):

ν = 2 − 2ρi,j + 12[SR2

i + SR2j − 2SRiSRjρ

2i,j

](15)

where ρ = σi,j

σiσjis the correlation coefficient of the excess returns of portfolio i and

j.To test the hypothesis, estimators for the Sharpe ratio have to be used.

SRi = mi

si

(16)

with

mi = 1T

T∑t=1

dit

si =

√√√√ 1T

T∑t=1

(dit − m2i

(Continued on next page)

66

dit = (Rit − Rbt)

where Ri and Rb are the returns on portfolio i and benchmark b, respectively.Calculating equation 15 with the estimators yields ν and as such the hypothesis

can be tested.The results depicted in table A1 as well as table A2 paint a clear picture.

Table A1: Sharpe ratio significance test results as suggested by Jobson and Korkie (1981)and Memmel (2003).

Year PF DMhigh PF DMlow z scores p-values

Sharpe ratio2013 -0.13 1.28 -3.478 0.0002014 1.56 0.97 1.320 0.1872015 0.15 0.92 -1.775 0.0752016 1.12 0.86 1.285 0.199All periods (2013-2016) 0.61 0.99 -0.622 0.534

This table shows the yearly Sharpe ratios of the DMhigh and DMlow portfolios aswell as the corresponding z-scores and probability values.

Table A2: Sharpe ratio significance test results as suggested by Jobson and Korkie (1981)and Memmel (2003).

Year PF DMhigh PF DMmiddle z scores p-values

Sharpe ratio2013 -0.13 1.39 -3.499 0.0002014 1.56 1.01 1.270 0.2042015 0.15 0.55 -1.095 0.2742016 1.12 1.24 0.237 0.813All periods (2013-2016) 0.61 0.24 -0.921 0.357

This table shows the yearly Sharpe ratios of the DMhigh and DMmiddle portfolios aswell as the corresponding z-scores and probability values.

67

Appendix B

Table B1: Excess returns of monthly size-adjusted portfolios within the portfolio analysiswith high (DMH), middle (DMM ) and low (DML) digital maturity portfolios. Excess returnscalculated with the Swedish one month Treasury bill rate.

Date PF DMH PF DMM PF DML DMH vs DML

Monthly excess returns in percent01/2013 2.98 6.78 3.15 -0.1702/2013 -1.98 9.52 5.80 -7.7803/2013 1.39 -3.61 1.41 -0.0204/2013 -1.10 -1.01 1.87 -2.9705/2013 2.06 3.40 3.46 -1.4006/2013 -8.18 -4.20 -4.60 -3.5807/2013 6.10 2.38 6.02 0.0808/2013 -2.70 1.80 -1.80 -0.9009/2013 -0.63 2.07 5.80 -6.4310/2013 1.08 1.42 -4.80 5.8811/2013 0.97 0.54 2.59 -1.6212/2013 -1.59 -0.31 -1.79 0.2001/2014 -2.26 0.59 -1.63 -0.6302/2014 6.92 5.10 2.04 4.8803/2014 -1.57 0.96 0.83 -2.4004/2014 2.10 6.09 -0.42 2.5205/2014 0.50 -0.24 3.75 -3.2506/2014 -2.45 -2.13 -2.80 0.3507/2014 1.68 -0.11 1.22 0.4608/2014 0.47 -3.75 -0.13 0.6009/2014 1.30 -3.23 -2.48 3.7810/2014 6.39 2.59 -1.82 8.2111/2014 1.74 1.32 5.73 -3.99

Continued on next page

68

Table B1 – continued from previous pageDate DMH DMM DML DMH vs DML

12/2014 0.78 3.99 5.28 -4.5001/2015 4.53 3.81 5.64 -1.1102/2015 3.38 10.67 8.26 -4.8803/2015 -1.45 1.69 2.51 -3.9604/2015 3.10 0.66 -5.96 9.0605/2015 -1.04 3.07 3.03 -4.0706/2015 -3.08 -6.55 -7.41 4.3307/2015 1.12 2.99 5.04 -3.9208/2015 -2.84 -7.36 -5.47 2.6309/2015 -6.88 -4.42 -0.47 -6.4110/2015 4.51 5.90 7.22 -2.7111/2015 2.70 5.12 10.19 -7.4912/2015 -2.16 -4.74 -3.17 1.0101/2016 -7.61 -5.24 -6.48 -1.1302/2016 1.69 3.04 0.57 1.1203/2016 5.38 5.04 7.12 -1.7404/2016 1.28 0.21 -3.93 5.2105/2016 2.24 0.73 3.68 -1.4406/2016 -4.27 -7.21 -3.47 -0.8007/2016 3.17 9.53 5.48 -2.3108/2016 1.69 6.63 2.75 -1.0609/2016 4.25 2.05 2.48 1.7710/2016 2.87 2.65 -6.09 8.9611/2016 -1.27 -1.71 5.66 -6.9312/2016 5.35 4.68 7.18 -1.83

69

Appendix C

Modigliani-Modigliani Risk-Adjusted Performance

Another way to measure and evaluate portfolio performance is suggested by Modiglianiand Modigliani (1997). This is an alternative measure of risk-adjusted performance.

RAPi =(

σM

σi

)(ri − rf ) + rf (17)

where RAPi is the risk-adjusted performance of portfolio i, σM is the standarddeviation of market returns, σi is the standard deviation of returns for portfolio i,rf the average return for portfolio i and rf the risk-free return.

Reformulating the risk-adjusted performance measure to solely base it on excessreturns (RAPA) yields (Farah, 2002):

RAPi − rf =(

σM

σi

)(ri − rf ) (18)

and henceRAPAi = σM

(ri − rf

σi

)(19)

Table C1 shows the results of the RAPA criteria for the DM portfolios.

Table C1: Result of the Modigliani and Modigliani (1997) risk-adjustedperformance measure (RAPA).

Year PF DMhigh PF DMmiddle PF DMlow

RAPA criteria2013 -0.14 % 1.48 % 1.36 %2014 1.43 % 0.92 % 0.89 %2015 0.16 % 0.62 % 1.03 %2016 1.25 % 1.39 % 0.96 %

70

Appendix D

Table D1: Digital maturity and subsequent operating performance panel regressionoverview.

Dependent Variable: ROAt+1Panel Generalized Method of MomentsSample Period: 2010-2015 (adjusted)Periods included: 6Cross-sections included: 24Effects Specification: Period fixedWhite period standard errors & covarianceVariable Coefficient Std. Error t-Statistic Prob.

ROA 0.652 0.085 7.646 0.000Intercept 9.223 2.628 3.509 0.001ln(CapEx/TA) -1.758 9.569 -0.183 0.855DM -0.880 0.438 -2.008 0.047∆ROA -0.388 0.031 -12.543 0.000ln(MV ) -0.132 0.059 -2.230 0.028Industry Dummy -1.161 0.417 -2.788 0.006

R-squared 0.525 Mean dependent var 5.880Adjusted R-squared 0.476 S.D. dependent var 5.330S.E. of regression 3.857 Sum squared resid 1577.2J-statistic 0.001 Instrument rank 12Durbin-Watson stat 2.156

This table reports the detailed results of the dynamic panel data regression estimatedwith Generalized Method of Moments and the Arellano and Bond (1991) estimator.

71